<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Hugh Evans on Medium]]></title>
        <description><![CDATA[Stories by Hugh Evans on Medium]]></description>
        <link>https://medium.com/@hugh.evans.dev?source=rss-d8aeb74e08e0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*EW28XYsf9Rljs41fN8g6Hg.jpeg</url>
            <title>Stories by Hugh Evans on Medium</title>
            <link>https://medium.com/@hugh.evans.dev?source=rss-d8aeb74e08e0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 11:49:09 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@hugh.evans.dev/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Making a QR Code Quilt, Again]]></title>
            <link>https://medium.com/@hugh.evans.dev/making-a-qr-code-quilt-again-ddb6ac4b52ff?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/ddb6ac4b52ff</guid>
            <category><![CDATA[digital-making]]></category>
            <category><![CDATA[qr-code]]></category>
            <category><![CDATA[sewing]]></category>
            <category><![CDATA[making]]></category>
            <category><![CDATA[quilting]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Tue, 15 Sep 2026 14:00:24 GMT</pubDate>
            <atom:updated>2026-09-15T14:00:24.414Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eJeSREsyPzwgRLppk-qVBg.jpeg" /></figure><p>I really like projects that explore the overlap of traditional crafts and computing / digital technology. Some of my favourite examples of this include <a href="https://bsky.app/profile/studioknotunknot.bsky.social/post/3mixwa2hzis2m">Madeleine Shepard’s hacked knitting machine cellular automata</a>, <a href="https://www.mozillafoundation.org/en/nothing-personal/anti-surveillance-fashion-privacy-ai/?gad_source=1&amp;gad_campaignid=23357096868&amp;gbraid=0AAAAADy9_2pW5v9wZn58ear-peSxBtCv0&amp;gclid=Cj0KCQjwh4TVBhCWARIsAG0czmp5fgzVZkPxpvwWoP-aBMOG7hiMRjGEYmqdpedHJrMFsJ-kt34U_nIaApAEEALw_wcB">adversarial print that protects against surveillance cameras</a>, and <a href="https://hachyderm.io/@dinobrarian@mastodon.social/117237225522923315">Charlie’s quiltagon project</a>. Coming back from EMF camp (the most traditional craft x digital technology place in the world) I was inspired to revisit a past project of mine exploring these themes for last year and improve on it.</p><p>Last year I had a whole stack of old black and white conference T-shirts and whilst trying to think about what to do with them I thought “QR quilt?” <a href="https://hughevans.dev/everything-i-made-in-2025/#:~:text=looking%20for%20parts.-,QR%20Code%20Quilt,-This%20QR%20code">I had no idea what I was doing but I made it</a>, the “quilt” (or QRuilt, thanks Jack) actually scanned but only when stretched in a particular way. This was a “postage stamp” quilt where all the modules were one individual fabric block.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0rJHxAg-uThO7oQRrJCGNg.jpeg" /></figure><p>I later learned that what I made above isn’t even really a quilt, more of a fancy cushion cover. A quilt comprises three layers: the front, batting which adds weight and warmth, and the back. Quilting is actually the process of sewing through all three of these layers to join them and ensure the batting is distributed evenly.</p><p>Facing down another stack of old conference shirts I decided to try making this project again, but this time I asked for some advice from my quilting-expert nan: who armed me with a stack of books, quilting rulers, adhesive spray, and curved safety pins as well as some encouragement throughout the project! Here’s how it went.</p><h3>Making a Plan</h3><p>Rather than making a postage stamp quilt like last time I opted for a simpler approach this time around: 25 8x8” white blocks with the black QR code modules appliqued on top. These would then be pieced together to form the QR code with a 2” white border added to avoid any of the black modules being folded over the edge like in my first version. I wasn’t really sure how I was going to finish the edge when I started, but eventually landed on creating a binding out of leftover fabric.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/907/1*l0LJvlE7Ffvdsty8PZw6-Q.png" /></figure><p>I kept a copy of the design of my QR code split into all 25 blocks to hand at all times throughout the project so I could keep track of where I was and make sure that I was putting my blocks together correctly.</p><h3>Cutting and Assembling Blocks</h3><p>T-shirt material is not ideal for quilting because as a knitted material it is very stretchy and prone to puckering and warping so after cutting all my (freshly washed) old t-shirts into usable sections I ironed on some interfacing to stabilise the fabric. The interfacing provided an interesting opportunity for this project in particular: for the white fabric I just used a simple one-sided interfacing, but for the black fabric I used a double sided, paper-backed, interfacing which had two main advantages. The paper backing helped provide extra stability to the black fabric whilst I was cutting it, and when it came to attaching the black details the fabric could be placed and pressed to fuse them into place before being appliquéd on my sewing machine.</p><p>I cut all the white blocks out with an 8” quilting square and rotary cutter on a large cutting mat. I did the same for the black details, and then used my Silhouette Cameo 5 vinyl cutter to cut out the black shapes using SVGs exported from my design in the previous section, re-using as much of the waste pieces as possible to cut down on fabric use. The vinyl cutter did a really good job of cutting through the black fabric with the fabric face down on a sticky cutting mat (with the interfacing backing paper left on), a depth of 7, force of around 32, and all other settings set to the default for cutting cotton fabric with the default knife attachment. I had a few hairy moments when the fabric came unstuck from the mat as unfortunately cutting fabric quickly removes the stickiness from the mat due to the dust produced, so I ended up using three old cutting mats whilst cutting all the black details.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qzTYCpan1E3XrQ6UG36i9A.jpeg" /></figure><p>After ironing all of the black details onto the white blocks I used black thread and a running stitch on my sewing machine to trace the edge of all the black details to make sure they were properly attached. After repeating this process for all 25 blocks I was ready to start piecing them together.</p><h3>Piecing Blocks</h3><p>Piecing these blocks was pretty simple as they were all square and the same size. I pinned the blocks together making sure to line up black modules neatly, used a running stitch to attach the blocks together, and pressed the seams flat as I went to keep things nice and square.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LLgG5wJHTkvFiofHB5yLXQ.jpeg" /></figure><p>Once I’d completed two rows I trimmed the corner of the end of each seam to remove excess fabric where the corners of four blocks met and then used one long seam to join each row together, repeating this process until I had a finished QR code!</p><p>Despite my best efforts there were some extremely visible but small gaps between the black modules around the center of the QR code, towards the end of the project I used a washable permanent marker to go over them and some errant white thread from quilting and it left the quilt looking much sharper!</p><h3>Adding the Border</h3><p>To add the border I cut 2” wide 8” long strips of the white, interfacing backed T-shirt fabric and sewed them together along the ends to create two 40” long rectangles and two 52” long rectangles (much longer than I needed but meant I could cut all my strips the same and use the rougher / stained ones at the ends). I pinned the shorter strips along the top and bottom of the pieced QR code and then attached them with a straight stitch, repeating the process for the two remaining sides with the longer strips and trimming off the excess with a quilting square to get some nice tidy corners for my finished front.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VIb1-M5FQen-Rqkrn5VwPg.jpeg" /></figure><h3>Making the Quilt Sandwich</h3><p>Next came making the back and cutting the batting, I made a huge rookie error here: I cut the back and the batting to the exact size of the front. This is a bad idea for several reasons but primarily because as you quilt the front it spreads out more than the batting and back, even when really well basted, which meant the edges of my finished quilt were missing some batting and there were gaps between the back and the binding in places. For better results I should have left 4” or so of extra material around the edge, this would also have given me the option to finish the edge of my quilt by folding over the back rather than making a binding. Otherwise this step went pretty smoothly, I made the back by sewing together two large pieces of cream cotton fabric and pressing the seam flat.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sSM_EA3joIHgf0iiqSkr9w.jpeg" /></figure><p>I basted all three layers together using curved safety pins. This took a while but worked pretty good, I think I may have got better results using the spray adhesive to baste the layers together instead as I definitely had some squishing and stretching in my blocks whilst quilting, but overall I was still happy with the result.</p><h3>Quilting</h3><p>This was the part of the project I was most nervous about, stitching through the whole sandwich to lock the layers in place. My nan suggested I try “stitch in the ditch” a quilting technique where you sew on top of the existing seams to hide your quilt on the front side. This was the hardest part of the project by far, and surprisingly physical even for this relatively small quilt. I replaced my sewing machine foot with a walking foot to help reduce the sliding of the quilt front. I sewed around the center square and worked my way out to evenly align the blocks. I did this in bursts and took several breaks, I also had some headaches where the curved safety pins got caught on my walking sewing foot because I’d placed them too close to the seams.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JUA2VeqH95O2bbTESOm36g.jpeg" /></figure><p>I managed to keep my stitches in the existing seams for the most part but some of my wonky piecing from earlier steps made this tricky. My less than steady hands also led to some slips and wonky stitches in places, I also had puckering throughout, particularly on the thicker sections where the seams joining the blocks added bulk, but not so much that I went back to redo it.</p><h3>Making and adding binding</h3><p>I made 3 metres of binding and it was exactly the right length which made attaching the last side a little nerve racking. I made this by cutting a series of 3” wide strips out of white t-shirt fabric with interfacing left over from making the blocks and sewing them together. I folded it in half and pressed it and then folded each side into the middle and pressed it again to create my finished binding. I found making the binding weirdly relaxing although it was definitely a little uneven in place and definitely too thick which became an issue in the corners.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dAdBZ1IrkRu6Y4zm6lx2TQ.jpeg" /></figure><p>When it came to attaching the binding I just pinned it to the raw edge of the quilted fabric and then sewed through the binding, front, batting, back, binding sandwich with a straight stitch. At the corners I folded the binding back on itself to create a neat corner but with the interfacing this meant the corners got pretty chunky. I worked my way around all four sides like this and was just short of binding to finish the last corner neatly so I just left it rough.</p><h3>Final product</h3><p>With the binding attached all that was left to do was to trim some excess threads and dust off my quilt. I am so happy with the result, not least that the QR code actually scans! Whilst it’s not quite big enough to use as an actual quilt it is warm and has a nice weight to it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pBI-1gcJZdrTpSwHygOviA.jpeg" /></figure><p>This project was so much better for asking for advice, taking the time to understand the techniques that make a proper-quilt, and generally at least aspiring to doing things “the right way” as I put it together as opposed to making things up as I go as I’m usually inclined to do.</p><p>I’m already excited to start my next quilting project, I really enjoyed putting this together and learnt so much for next time.</p><p>I’m toying with the idea of building a tool for automatically creating SVGs and a PDF pattern for QR quilts, if that’s something you’re interested in let me know.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ddb6ac4b52ff" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My PyData London 2026 Conference Highlights]]></title>
            <link>https://medium.com/@hugh.evans.dev/my-pydata-london-2026-conference-highlights-52ebcc85dd8b?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/52ebcc85dd8b</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[community]]></category>
            <category><![CDATA[conference]]></category>
            <category><![CDATA[pydata]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Thu, 11 Jun 2026 11:09:44 GMT</pubDate>
            <atom:updated>2026-06-11T11:10:35.614Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*reEj--w5DB5SVkL0TM_cXw.jpeg" /></figure><p>PyData London is a conference bringing together data scientists and engineers from all over the world for workshops and talks covering everything from document parsing to weather forecasting. This was my fourth year volunteering and I was really excited for the conference, not least because Aiven would be sponsoring this year which was a great opportunity to give back to a community that means so much to me.</p><p><em>For the full, free, and up to date version of this blog and more like it </em><a href="https://hughevans.dev/my-pydata-london-2026-conference-highlights/"><em>check out the original version on my website.</em></a></p><p>I went into the conference hoping to learn something and catch up with friends old and new, as well as seeing some of the folks I met on <a href="https://hughevans.dev/the-tour-de-pydata-challenging-myself-to-speak-at-every-pydata-meetup-in-the-uk-and-ireland/">my tour of UK PyData groups</a>. Now that the dust has settled (and I’ve had some time to recover from a packed weekend) I’ve written up my highlights from the event.</p><h3>Friday (Workshops)</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*CLn608A0Lwjx17JTP48QhA.jpeg" /></figure><p>After arriving bright and early with a cab full of last year’s conference supplies (97 PyData rubber ducks anyone?) I settled in for a day of workshops. It was difficult to pick between the three tracks but I ended up going to the Docling workshop, the MCP OTel workshop, and Cheuk’s workshop about LightEval — all of which were fantastic, although the Tom Nook “Tanuki” MCP from Fei and Tun’s workshop definitely stole my heart and Cheuk’s workshop was by far the most challenging.</p><p>We rounded out the day with the social at Fleets I (badly) organised (mentioning that there was a tab before we arrived would have been smart, sorry John S and John C who kindly sponsored)!</p><h3>Saturday</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*1E9RD-3ZnMNatxS2QWOYCQ.jpeg" /></figure><p>Saturday kicked off with an amazing keynote from Rachel-Lee Nabors which was a much needed story about the winding paths careers can take, the importance of community, and hope in a time of layoffs. I’m extremely grateful to Rachel-Lee for speaking, I can’t overstate the positive impact this talk had and I know I wasn’t alone in coming away inspired and hopeful.</p><p>I spent the rest of the day on the Aiven booth having conversations about our work on Diskless Kafka and our free and developer tiers. I got to chatting with MK from NumFOCUS about my PyData maps project and his awesome <a href="https://www.youtube.com/playlist?list=PLGVZCDnMOq0oHIJN8Hj3MgvpVzW8oTpf4">Give Me 5 series</a>.</p><p>It seemed like by Saturday (almost) everyone had figured out the badge maker Aiven sponsored, and every time I walked past I could see people making a badge which was really cool. Likewise <a href="https://hughevans.dev/i-built-an-arcade-game-things-went-wrong/">our Disk Invaders arcade cabinet</a> definitely captured people’s attention with no fewer than 82 games played over the weekend.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FGweh3nFE5Wk%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fshorts%2FGweh3nFE5Wk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FGweh3nFE5Wk%2Fhq2.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="640" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/3323ea22df220f0d756051e1b72fcad6/href">https://medium.com/media/3323ea22df220f0d756051e1b72fcad6/href</a></iframe><p>I was also interviewed by Cheuk along with some friends for her YouTube channel about our experiences at PyData London. We wrapped up the day with a social event in the venue and then I grabbed some dinner with the other organisers nearby.</p><h3>Sunday</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*nXIpQ_dR1B5gVKGMLhq3lw.jpeg" /></figure><p>I had the pleasure of MCing the lightning talks Sunday morning which was a real treat and we had some fantastic speakers, all of whom kept to the 5 minute time limit like absolute pros. We didn’t make it through everyone who signed up but everyone who didn’t get a chance to speak has been <a href="https://london.pydata.org/submit-a-talk/">invited to speak</a> at the PyData London Meetup.</p><p>At lunch the PyData organisers meetup quickly ran out of chairs and we ended up hunting for more. There were some great opportunities and having 30+ UK PyData organisers all in one place meant it was easy to work through the problems some organisers are facing. It was also a great reminder of just how big the UK PyData community is.</p><p>After lunch I went along to Damian’s talk exploring the correlation between NIMBYism and cancelled renewable energy projects. This was my third time seeing this talk and I enjoy it more every time! The last talk of the event I saw was Adam Hill’s talk on creating quizzes based on PDFs with LLMs as judges.</p><h3>Looking Forward to 2027</h3><p>As ever this year the volunteers were amazing and put on a fantastic event that I was proud to be a part of. I’m really looking forward to next year’s event. The conference closed out with the volunteer dinner, where we celebrated another successful event and the bittersweet end of John Carney’s time as conference chair. John is a lovely guy and has done a fantastic job these past few years, but I know we’re in safe hands with Prashant and Jiarui as our chairs for 2027.</p><p>See you at the next one! And if you want some PyData in the meantime, why not <a href="https://hughevans.dev/pydata-maps/#:~:text=may%20be%20older!)-,Classic%20map,-Plain%20markers%20for">check out your local group</a>?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=52ebcc85dd8b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Tour de PyData: Challenging Myself to Speak at Every PyData Meetup in the UK and Ireland]]></title>
            <link>https://medium.com/@hugh.evans.dev/the-tour-de-pydata-challenging-myself-to-speak-at-every-pydata-meetup-in-the-uk-and-ireland-a9c752ed8e8f?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/a9c752ed8e8f</guid>
            <category><![CDATA[web-scraping]]></category>
            <category><![CDATA[mapping]]></category>
            <category><![CDATA[community]]></category>
            <category><![CDATA[pydata]]></category>
            <category><![CDATA[python]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Wed, 25 Mar 2026 11:29:30 GMT</pubDate>
            <atom:updated>2026-03-25T13:04:28.936Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Ck-MhMVN8lDS44Ec.png" /></figure><p>I have a habit of setting myself ambitious if slightly wacky goals: riding the London to Brighton bike ride, <a href="https://github.com/hevansDev/ascii-booth?ref=hughevans.dev">building an ASCII art photobooth</a>, pursuing a career in dev rel etc and my latest is no exception. One of my aims for 2026 was to <a href="https://hughevans.dev/everything-i-made-in-2025/#:~:text=On%20the%20subject,as%20I%20can.">give back to the Python community</a> and to that end I’m helping to organise meetups for PyData London, running socials at this year’s PyData London conference (literally organising a piss up in a brewery) but I wanted to do more.</p><p><em>For the full, free, and up to date version of this blog and more like it </em><a href="https://hughevans.dev/the-tour-de-pydata-challenging-myself-to-speak-at-every-pydata-meetup-in-the-uk-and-ireland/"><em>check out the original version on my website.</em></a></p><blockquote><a href="https://pydata.org/?ref=hughevans.dev">PyData</a>, for those who are unaware, is the educational/community arm of <a href="https://numfocus.org/?ref=hughevans.dev">NUMFocus</a>, an organisation that supports many open-source scientific and data science Python projects like <a href="https://scipy.org/?ref=hughevans.dev">Scipy</a> and <a href="https://pandas.pydata.org/?ref=hughevans.dev">Pandas</a>. PyData has groups all over the world hosting regular meetups on a variety of topics.</blockquote><p><a href="https://hughevans.dev/mapping-the-pydata-community-2/">As I covered in my blog post on the topic</a> I’ve long been equally inspired and frustrated by the map of PyData Meetups: frustrated by its inaccuracies (and, as I’ll dig into, incompleteness), and <a href="https://hughevans.dev/everything-i-made-in-2025/#:~:text=On%20the%20subject,as%20I%20can.">inspired to visit some of those groups</a>. After a few months of careful preparation, tentatively reaching out to other PyData organisers across the UK, and ultimately <a href="https://hughevans.dev/pydata-maps/">making my own map(s) of PyData groups</a> I was ready to begin my journey.</p><h3>The Tour de PyData</h3><p>So here’s the plan: I’m going to try and speak at as many PyData Meetups across the UK and Ireland as possible this year. My plan for this little adventure is both to see some places I’ve never visited before (after two years in dev rel I’ve been to Charlotte airport more than I’ve been to Scotland) and to help promote something really cool: across the UK and Ireland are more PyData groups than anywhere else in the world.</p><p>I’ll be documenting my journey here and doing my best to showcase the depth and breadth of the wonderful PyData community.</p><p>What better time to call attention to our awesome community than in 2026 with community budgets tightening across the board and volunteers needed more than ever?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iZ9uAP4pl0UPMs6DXXzZlg.png" /></figure><p>Above is yet another map I cooked up, this one listing all the PyData groups I’m scheduled to speak at and charting my progress. To make this challenge even close to possible I’m limiting the list of groups I plan to speak at to groups that are regularly hosting events, bringing the total number of groups down from 21 to 15. A huge thank you to all the PyData organisers who have invited me to speak at their groups already!</p><h3>Another Problem With The Meetup Map</h3><p>When I last spoke about this project I outlined four major issues I had with <a href="https://www.meetup.com/pro/pydata/?ref=hughevans.dev">the official PyData group map on Meetup</a>: it’s impossible to see how many meetups are in the same city, several groups are in the wrong location, it’s unclear if groups are still active, and every group is shown dozen of miles north of its actual location. After the talk <a href="https://www.linkedin.com/in/stelios-christodoulou-4869aa250/?ref=hughevans.dev">Stelios Christodoulou</a> raised <a href="https://github.com/hevansDev/mapping-pydata/issues/1?ref=hughevans.dev">an issue</a> on my GitHub repo for the project that would expose a fifth, even more significant, problem. Stelios pointed out I was missing <a href="https://www.meetup.com/pydata-edinburgh/?ref=hughevans.dev">PyData Edinburgh</a> from my map. Initially I kicked myself, wondering what I’d missed in my webscraping code that had caused me to miss a group as large as Edinburgh, before realising this was in fact a symptom of a wider problem: <strong><em>not every PyData group is included in the official map!</em></strong></p><p>I was able to write <a href="https://github.com/hevansDev/mapping-pydata/tree/main/experiments/ScanMeetupByCity?ref=hughevans.dev">a hacky script</a> to run through every major city on Earth looking for PyData groups and turned up some interesting results:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OGB5AniA_PP4NUH5k7xuMw.png" /></figure><p>7 PyData groups are missing from the official map: Tokyo, Abu Dhabi, Vilnius, Krakow, Poznan, Basel, Lausanne, and of course, Edinburgh. Why are these groups missing from the official map? It’s unclear but all of these groups have been inactive for a while, which might explain why. My search also turned up some interesting <em>former</em> PyData groups like <a href="https://www.meetup.com/pydata-lancaster/?ref=hughevans.dev">Software Talks Lancaster</a>, which still have “PyData” group URLs but seem to have moved onto new topics.</p><h3>Grand Depart</h3><p>I’m set to get started on my Tour de PyData later this week: starting off with the wonderful PyData Manchester, the UK’s second largest PyData group, and PyData Hull, the newest. Of course the obvious question is what do I plan on speaking about 15 times? The PyData mapping project! A little meta though it may be this project is a great fit for PyData events, as it features: data engineering, web scraping, geoencoding, and of course Python. Now, with my map of PyData groups more complete than ever, I’m excited to share it with the community.</p><p>I’m really looking forward to spending some time with the PyData community this year, and hopefully to help encourage some folks to support their local PyData group. Groups always need support in the form of speakers, organisers, venues, sponsors, and in general just people to show up and build the community.</p><p>Are you up for the challenge of speaking at every PyData Group in the UK and Ireland in a year? Feel free to <a href="https://github.com/hevansDev/mapping-pydata/tree/main/experiments/TourDePyData?ref=hughevans.dev">fork my map</a> and create your own if you like.</p><p>I’ll be updating on the progress of my journey here, watch this space and see you at a local PyData soon!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a9c752ed8e8f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Securing My Homelab With Tailscale]]></title>
            <link>https://medium.com/@hugh.evans.dev/securing-my-homelab-with-tailscale-b6835268e9cd?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/b6835268e9cd</guid>
            <category><![CDATA[reddit]]></category>
            <category><![CDATA[tailscale]]></category>
            <category><![CDATA[homelab]]></category>
            <category><![CDATA[security]]></category>
            <category><![CDATA[youtube]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Sun, 22 Mar 2026 19:23:55 GMT</pubDate>
            <atom:updated>2026-03-22T19:23:55.332Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pN9c54fn6X0BLaIh66k6ng.png" /></figure><p>Follow up video <a href="https://hughevans.dev/a-homelab-cautionary-tale-how-crypto-scammers-hacked-my-analytics-dashboard/">to my blog on my homelab getting hacked</a>, covering how I secured Umami with Tailscale.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FEYA5skoiGQA%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DEYA5skoiGQA&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FEYA5skoiGQA%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/f5791d563d414cd8fdf08674a1f55ae9/href">https://medium.com/media/f5791d563d414cd8fdf08674a1f55ae9/href</a></iframe><p><em>For the full, free, and up to date version of this blog and more like it </em><a href="https://hughevans.dev/securing-my-homelab-with-tailscale/"><em>check out the original version on my website</em></a><em>.</em></p><p>Photo credit <a href="https://www.linkedin.com/in/ellie-geddis/?ref=hughevans.dev">Ellie Geddis</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b6835268e9cd" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A Homelab Cautionary Tale: How Crypto Scammers Hacked My Analytics Dashboard]]></title>
            <link>https://medium.com/@hugh.evans.dev/a-homelab-cautionary-tale-how-crypto-scammers-hacked-my-analytics-dashboard-616520be9129?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/616520be9129</guid>
            <category><![CDATA[web-analytics]]></category>
            <category><![CDATA[umami]]></category>
            <category><![CDATA[homelab]]></category>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Mon, 16 Mar 2026 11:24:04 GMT</pubDate>
            <atom:updated>2026-03-16T11:24:04.118Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*t4nD9fFqCjp6vYoJ.png" /></figure><p>I’ve been into homelabbing for a few years now. For those unaware, a homelab is a self managed IT environment, typically used for hosting apps and learning, which in practice usually means a hodgepodge of compute, storage, and networking.</p><p><em>For the full, free, and up to date version of this blog and more like it check out the </em><a href="https://hughevans.dev/a-homelab-cautionary-tale-how-crypto-scammers-hacked-my-analytics-dashboard/"><em>original version on my website</em></a><em>.</em></p><p>Here’s my homelab:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*b_RcziRyzQAYfuTo.png" /></figure><p>Here’s the parts of my homelab actually plugged in and doing meaningful(ish) work:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*hs6wTM7V2acBEyxE.png" /></figure><p>In my defence the desktop is the only Windows box in my flat and I’m damned if I’m unracking that UPS, even if the batteries have long since ceased to hold charge.</p><p>The three clustered Raspberry Pi 4s provide me with a decent enough environment to play around. I use my homelab to self-host things like my website, HomeAssistant, personal projects, and analytics to see if anyone is using any of the former. I had fun gradually bodging this together over the years until eventually my lacklustre approach to system administration best practices came back to bite me.</p><h3>Discovering the Hack</h3><p>Cryptocurrency and all other financial instruments with the crypto prefix are in my view at best ponzi schemes and legal sleight of hand to dodge gambling regulations and at worst climate destroying instruments of global corruption and crime. I’d thoroughly recommend <a href="https://www.mollywhite.net/?ref=hughevans.dev">Molly White’s</a> writing if you’re interested in what a mess the crypto industry is.</p><p>With all that said, you could imagine I was particularly annoyed when checking my web analytics dashboard to find that not only had someone hacked my homelab: they had somehow managed to inject a crypto casino ad into the web app html. I didn’t actually get a screenshot of this but it looked a little something like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*UdzSB1d4qGeZ27ZL.png" /></figure><p>After recovering from the digital equivalent of finding a stranger in my living room, I started digging into how exactly this happened.</p><h3>The Vulnerability</h3><p>A quick web search turned up<a href="https://github.com/umami-software/umami/issues/3852?ref=hughevans.dev"> an issue on GitHub describing exactly what I was seeing</a>. In short, a vulnerability in<a href="http://next.js/?ref=hughevans.dev"> Next.js</a> (<a href="https://nextjs.org/blog/CVE-2025-29927?ref=hughevans.dev">CVE-2025–29927</a>) allowed attackers to access compromised systems with Umami installed. Two things became immediately clear: one, that I had been very lucky, the scope of the impact on my homelab was very limited; two, I was late to the party–I hadn’t noticed this issue until late February 2026, despite the vulnerability having been known about since December 2024.</p><h3>Assessing the Blast Radius</h3><p>I deployed apps to my homelab with Docker, which limited the attacker’s access to the Umami container. Whilst the attacker gained authenticated access to my Umami dashboard, which allowed them to insert the gambling popups visible to anyone viewing that dashboard, the host server was not compromised. After stopping the container, and carefully inspecting its contents, I could see that the attacker had injected a malicious<a href="http://middleware.ts/?ref=hughevans.dev"> middleware.ts</a> file into the file system. This file seemingly would have caused more serious issues, but luckily the injection was partially blocked by file permission restrictions on the container. I’d like to chalk this up to me implementing <strong><em>robust zero-trust across best practices™</em></strong> across my entire homelab but this was mostly dumb luck.</p><p>After figuring out how the attack had compromised my Umami container I stopped it and did some digging around in the filesystem to understand the impact. Having seen in the GithHub issue evidence of malware more serious than just adding ads to my dashboard, I gave the host a thorough check to make sure it hadn’t been compromised: I looked for evidence of this in the form of new shell profiles, cron jobs, and any newly created files.</p><p>All in all, after a tense couple of hours spent inspecting my homelab I was able to confirm that the hosts and other containers were completely unaffected. The analytics data in the Umami database showed no evidence of tampering, and visitor data across my personal projects had not been accessed.</p><h3>Remediation</h3><p>Thanks to Docker limiting the spread of the attack, remediation was fairly simple. I deleted the compromised container, rotated my Umami database credentials, and pulled the patched image:</p><pre>docker rm umami<br>docker image pull umami</pre><p>I then recreated the container with the direct port 3000 exposure removed from my Docker Compose file and updated the admin password.</p><p>I also checked in on all my other containers for any other outstanding CVEs and updated them across the board.</p><h3>Lessons Learned</h3><p>Despite having spent years being the “<a href="https://jonathan.bergknoff.com/journal/always-pin-your-versions/?ref=hughevans.dev">pin your versions</a>” guy at work, I was running containers on :latest tags with no update monitoring, which led to being five months behind on the patch that would have protected me from this issue. Whilst Docker container isolation was the key factor that prevented this from being a full server compromise, I learned that I need to be more careful with my homelab security: no more default passwords for me!</p><p>Some next steps to further secure my homelab are adding a basic nginx authentication layer in front of my Umami dashboard and other private services, and setting up container update monitoring (potentially with<a href="https://awesome-docker-compose.com/diun?ref=hughevans.dev"> Diun</a>, which looks promising) to avoid missing another CVE for the best part of six months.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=616520be9129" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Mapping the PyData Community]]></title>
            <link>https://medium.com/@hugh.evans.dev/mapping-the-pydata-community-a5a264379f2a?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/a5a264379f2a</guid>
            <category><![CDATA[community]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[maps]]></category>
            <category><![CDATA[meetup]]></category>
            <category><![CDATA[pydata]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Thu, 26 Feb 2026 10:23:05 GMT</pubDate>
            <atom:updated>2026-02-26T10:23:05.212Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g-qmrHj5OtnBCUzu4Y_cSA.png" /></figure><h3>Gathering data, building maps, and the places that it can take you</h3><blockquote><em>For the full, free, and up to date version of this blog (interactive maps and all!) and more like it check out the </em><a href="https://hughevans.dev/mapping-the-pydata-community-2/"><em>original version on my website</em></a><em>.</em></blockquote><p><strong>I really like maps.</strong> I collect (mostly transit) maps and, when sitting down to write this blog, I quickly ran out of fingers counting all the maps up on the wall in my flat. Maps have an ability to inspire us to explore the world around us, within them containing just enough information to travel an unknown landscape whilst still leaving enough blank space to inspire the imagination. I also really like being a part of the PyData community having spent the past few years attending and later helping to organise PyData London.</p><p>PyData, for those who are unaware, is the educational/community arm of <a href="https://numfocus.org/?ref=hughevans.dev">NUMFocus</a>, an organisation that supports many open-source scientific and data science Python projects like Scipy and Pandas. PyData has groups all over the world hosting regular meetups and each PyData London meetup starts with a slide f<a href="https://www.meetup.com/en-AU/pro/pydata/?ref=hughevans.dev">eaturing this map of those communities</a>. Odds are if you’ve been to a PyData event before it was PyData London as with 16,000 members it’s the largest PyData community in the world.</p><p>PyData UK is much more than just a mega-meetup in the capital though, in fact the UK has more PyData groups than any other country in the world with a whopping 20 groups, narrowly beating out the US where there are 19. This abundance of groups means there’s likely a PyData community near you. Meetup’s search feature has long been a bit of a joke so if you want to find your nearest PyData community (or any Meetup community for that matter) your only real option is to use the map Meetup provides. And that map has a problem.</p><h3>Meetup’s Map is Terrible</h3><p>Despite its flaws I’ve always found this map particularly inspiring, much like <a href="http://pythondeadlin.es/?ref=hughevans.dev">pythondeadlin.es</a> and other similar resources, it showcases the sheer scale of the international Python community, and serves as something of an invitation to go seek out like-minded people the world over. For all that inspiration though, there’s a major issue with it: it’s built and maintained by Meetup. <a href="https://hughevans.dev/meetup-is-terrible-now-codebar-festival-fringe/">I’ve spent a lot of time complaining about Meetup over the last few years</a> (no hate to the folks actually building the Meetup platform, I’m sure it’s very difficult trying to get work done in the kind of environment private equity takeovers create). It’s a tool I rely on heavily as a community organiser, but if we zoom in specifically on the UK position of the map we can see the issues with Meetup’s map of PyData communities.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*aaNlDjT-68G7OboW.png" /></figure><p>Immediately obvious is this issue of how Meetup chose to display the locations of each group with every marker sitting dozens of miles north of the group’s actual location. Closer inspection reveals some more glaring issues: some groups like PyData Katsina are shown in entirely the wrong location (not just a few miles off the mark but in fact on the wrong continent), and it’s impossible to see if there is more than one group in the same city.Rather than fall back on my usual shtick of complaining about Meetup I decided to build my own improved version of this map. I was motivated to do this because of what this map, despite its flaws, inspired me to do which was to see how many UK Meetups I could actually speak at or attend: PyData London is an awesome community and the promise contained within this map is two dozen other communities of like minded people all over the UK. This particular goal required some data that was currently entirely absent from this map: which groups are actively hosting events and how do you submit a talk. I’ve previously spoken at PyData Southampton but this project would require more preparation than a friendly email to Adam Hill, so I set out to make my own maps of the PyData community that addressed the issues with the Meetup map and would provide me with the information I needed to plan my little adventure.</p><h3>Making My Own Maps</h3><h3>Collecting Group Data</h3><p>I had a couple of options for collecting the data I would need to produce my maps: use the Meetup API (which I have access to as a paid up Meetup Pro organiser), or scrape the data directly from the Meetup website (<a href="https://www.meetup.com/robots.txt?ref=hughevans.dev">as permitted by their robots.txt</a>). Meetup deprecated their REST API in 2025, <a href="https://github.com/search?q=meetup+api&amp;type=repositories&amp;ref=hughevans.dev">which in turn killed off various third-party wrappers</a>, and their docs now start with a <a href="https://www.meetup.com/graphql/?ref=hughevans.dev#:~:text=OAuth%20Clients-,Benefits%20of%20GraphQL,-GraphQL%20is%20a">“GraphQL good” screed</a> explaining their decision to do this. Apart from that somewhat jarring piece of product management, the Meetup API is actually fairly robust and revisiting it, I found that for getting groups from the PyData network you could easily use a GraphQL query like:</p><pre>query($cursor: String) {<br>          proNetwork(urlname: &quot;pydata&quot;) {<br>            groupsSearch(input: { first: 20, after: $cursor }) {<br>              totalCount<br>              pageInfo { endCursor hasNextPage }<br>              edges {<br>                node {<br>                  id<br>                  name<br>                  urlname<br>                  city<br>                  memberships { totalCount }<br>                }<br>              }<br>            }<br>          }<br>        }</pre><p>Granted this requires <a href="https://www.meetup.com/graphql/authentication/?ref=hughevans.dev">negotiating the Oauth process</a> and refreshing tokens periodically. I also particularly like their <a href="https://www.meetup.com/graphql/playground/?ref=hughevans.dev#graphQl-playground">GraphQL Playground</a> in the API Docs which uses your logged in Meetup account to give you access to a live GraphQL sandbox where you can experiment with queries, this is a nice touch and for me it’s a sign that there are actually engineers at Meetup who care about user experience.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*NJYWeWxcuIK9tvHb.png" /></figure><p>Ultimately though for simplicity I opted to just scrape everything I needed for my maps work from a local copy of the group data in a Pandas dataframe. Whilst the API does make it relatively easy to request some of the data I needed, like group names and upcoming events, it doesn’t have some of the data that was crucial for building maps of active groups, namely time since last event, and total number of past events.</p><p>Scraping from Meetup brings its own challenges. The list of groups on Pro Network pages is loaded dynamically. My code for scraping from the Meetup group list is sprawling and clunky (<a href="https://github.com/hevansDev/mapping-pydata/blob/main/PyDataMap.py?ref=hughevans.dev">you can check it out on GitHub here</a>) but boils down to loading the page with PlayWright, spoofing scrolling inputs and periodically dumping the static HTML to get a complete list of all the groups in the PyData Pro Network. I also used a similar approach to scrape the past and upcoming events for each individual group so I could collect the activity data I wanted.</p><h3>Data Preparation</h3><p>After scraping the data some preparation was needed to make it usable for mapping. The main issue (as I previously mentioned) is that the latitude and longitude for each group is incorrect when scraped from Meetup, and the city field is also junk because for some groups it was populated wrong i.e. PyData Katsina’s city field is set to Norwich. This meant that the only “accurate” location information for groups on Meetup is the group name itself. Most PyData groups are named something like “PyData &lt;location&gt;” i.e. PyData London, PyData Boston etc so I was able to use string replacement to remove common phrases like “PyData” to get just the city name string. I could then geoencode the city name to return plottable co-ordinates with <a href="https://github.com/geopy/geopy?ref=hughevans.dev">geopy</a>. This worked great apart from a few notable exceptions like PyData En Español Global and PyMc Global which don’t have physical locations and some groups which ignore the naming convention entirely, I’m looking at you “Datenanalyse, Data Science und Statistik — PyData Dortmund”.</p><pre>{<br>  &quot;hints&quot;: {<br>    &quot;PyMC Online Meetup&quot;: null,<br>    &quot;PyData En Espa\u00f1ol Global.&quot;: null,<br>    &quot;NEO AI - a PyData Group&quot;: &quot;Cleveland, Ohio, USA&quot;,<br>    &quot;PyData Ireland&quot;: &quot;Dublin, Ireland&quot;,<br>…</pre><p>I got around this by adding a list of “hints” for the geoencoder which allowed me to flag which online only groups should be ignored and create aliases for other groups where required. The only other data preparation required was a simple join of all group lists and individual group data (i.e. past event count, total members) which was easy enough with Pandas.</p><h3>Bonus: Playing with the Dataset</h3><p>One fun upshot of going to the effort of scraping, enriching, and preparing the PyData community data was that it allowed me to easily interrogate the data in a <a href="https://marimo.io/?ref=hughevans.dev">Marimo</a> notebook. You can clone the repo to play around with the data yourself but here are a few insights I found particularly interesting:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/632/1*Pm0cGBDTZoqKd78qPiE4kg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/473/1*6xdzI3ouWNETGPILBJHonA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/455/1*YW8bQd8thmT74bnRQW_mIg.png" /></figure><h3>Plotting Maps</h3><p>For plotting the maps I elected to use <a href="https://github.com/python-visualization/folium?ref=hughevans.dev">Folium</a>. Folium has its limitations but is well suited for easily creating custom markers to make the types of maps I wanted.</p><p>Based on data scraped from <a href="https://www.meetup.com/pro/pydata/?ref=hughevans.dev">www.meetup.com/pro/pydata</a>, last updated 25 February 2026 at 20:17 GMT (data for individual groups may be older!)</p><p>The first map I plotted is this one with plain markers for each PyData and PyData affiliated group closely mimicking the existing PyData group map on meetup. The only difference is the location of groups is plotted as the geocoded name.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/514/1*KpQqG59j_U9d2-7uEiBsag.png" /></figure><p>Making this second map was a decent chunk of the motivation for this entire project. My goal was to make it easy at a glance to spot active groups to speak at or attend, and to identify dormant or inactive groups that may require support (although I later decided to split that last part into the third map below). To that end I set up custom markers: the marker radius is scaled with the number of group members, the markers become fainter the longer it has been since the last event was scheduled, blue groups have no events scheduled, and green groups have an event scheduled. The blue / green distinction I added in later so brand new groups wouldn’t appear as faint but instead would be extremely prominent on the map, I added this feature after spotting the new PyData Hull group that was created during this project.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/514/1*mKvxjQ-Vx1A1vF9_ABmSkw.png" /></figure><p>This final map shows groups that haven’t hosted an event in 100 days in bright red. I decided to make it because the map above by design makes groups like these almost invisible and I felt it would be useful to specifically draw attention to groups that may need support. I landed on 100 days as even mostly dormant groups that still had active organisers would have uploaded the link to <a href="https://pydata.org/global2025?ref=hughevans.dev">last year’s PyData global</a>, which at time of writing was 77 days ago, meaning that it is likely that these groups are in desperate need of organisers or support in order to host events in future.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/517/1*qcDZuVzgJWcjTJb62P51mA.png" /></figure><h3>Publishing and Keeping Maps Up to Date</h3><p>Up until this point in the project I’d been hacking away in a Marimo notebook which took up to 15 minutes to scrape the data from Meetup and produce the maps. Before I shared the maps I wanted to make sure they would stay up to date with the data available from Meetup. To this end I created a Python script that would spit out the maps as static HTML that I could serve from GitHub pages and then set up a GitHub action to run the script every 24 hours. I also added some basic caching to handle edge cases when the scraping timed out. I also went back and updated the maps at this point to allow specific zooms and views to be saved in the URL to make it easy to create country or region specific maps.</p><p>I had a lot of fun with this project and these maps did end up being useful for thinking about a potential tour of UK PyData groups, even if it did mostly reveal the scale of that challenge. From a technical perspective I wouldn’t use Folium if I were to do this again because of limited options for maps in different languages, ideally I would like to create localised views of maps for each PyData group.As part of this project I presented my work as a lightning talk at PyData London, and subsequently the group used my map to replace the existing one (as it turned out I wasn’t the only one annoyed by the locations being wrong on the original meetup map) which was really cool, it’s always nice to see tools you build getting adopted.</p><p>I’m proud to be a part of the international PyData community, but this project has made me especially proud to be part of the UK PyData community: I’ve learned so much by coming along to PyData over the last few years and I know I’m not alone in that. The UK PyData community is an amazing resource that we’re lucky to have but it only exists because of the time and effort our community invests in it, especially the awesome organisers of these groups, some of whom I’ve been lucky enough to meet at the PyData London conference organisers lunch over the years.</p><p><strong>If you take one thing away from this blog: if your local PyData group hasn’t hosted an event in 100 days, please consider supporting them either through volunteering, speaking, sponsoring, or otherwise participating in building this awesome community. </strong>If you aren’t inspired by maps to seek out regional PyData communities, I’m sure meeting the organisers of these groups would, they’re great people!</p><p>Watch this space for updates on my journey to as many UK PyData meetups who will have me. If you’re an organiser and you’d like me to speak, give me a shout!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a5a264379f2a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Everything I made in 2025]]></title>
            <link>https://medium.com/@hugh.evans.dev/everything-i-made-in-2025-4357ee11aaba?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/4357ee11aaba</guid>
            <category><![CDATA[community]]></category>
            <category><![CDATA[devrel]]></category>
            <category><![CDATA[engineering]]></category>
            <category><![CDATA[year-in-review]]></category>
            <category><![CDATA[making]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Tue, 23 Dec 2025 18:48:25 GMT</pubDate>
            <atom:updated>2025-12-23T18:49:39.606Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CinpJ9h_bLtk4VtShKYFfw.jpeg" /></figure><p>Revisiting 2025 through projects made.</p><p>This year has been a lot like the year before it, I again started the year at one company before moving to a Developer Advocate role. I again spent October in the states working conferences, and again, took a picture in front of a bridge.</p><p><em>For the full, free, and up to date version of this blog and more like it check out the </em><a href="https://hughevans.dev/everything-i-made-in-2025/"><em>original version on my website</em></a><em>.</em></p><p>There were some pretty major differences though, not least that I didn’t get made redundant in October this time. In fact my role as a Developer Advocate at Aiven has been great and has given me the chance to do some great work with the Kafka community on <a href="https://hughevans.dev/beginners-guide-diskless-apache-kafka-kip-1150/">Diskless Kafka</a> and the <a href="https://getkafkanated.substack.com/?ref=hughevans.dev">Get Kafka-Nated podcast</a>.</p><p>One big change this year was putting a lot of my time and energy put into community work: I attended and spoke at more conferences this year than I had in the rest of my career, built a team of volunteers at AI Signals, got involved in organising <a href="https://www.meetup.com/pydata-london-meetup/?ref=hughevans.dev">PyData London monthly events</a> in addition to my role as a conference organiser, and set up the <a href="https://notanother.pizza/?ref=hughevans.dev">Not Another Pizza</a> community organiser group.</p><p>What hasn’t changed though is my drive to be making. Now that I am working in a senior role (and shooting for staff in the new year) my time to work on personal projects has been much less than I would like, but I still took every chance I could this year to be making and learning.Like last year I would like to revisit this year’s projects and share what I learned from each one. Read on for a year in making, explored one project at a time, and my thoughts for the year to come. This is a long one so I won’t be offended if you skim for bits that interest you or skip to the conclusion.</p><h3>“Artemide” Table Lamp</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*OtDhXNayezp80uOS.png" /></figure><p>The Artemide Nesso Table Lamp is an iconic piece of 1960s design that is much imitated. One such imitation bought on Temu stopped working days after arrival in my flat prompting me to upgrade it to get it up to scratch with the real thing.</p><p>The Temu version came with a flakey capacitive touch and LED setup that didn’t match the warm and cosy vibe of the original. As I often do, I looked at it and went “I can fix that” took it apart, got stuck figuring out what to replace the LEDs with, and ended up chucking it in a drawer. It wasn’t until <a href="https://www.tiktok.com/@wesleybreed/video/7312976193895075103?lang=en&amp;ref=hughevans.dev">I saw this video online</a> from <a href="https://www.tiktok.com/@wesleybreed?lang=en&amp;ref=hughevans.dev">wesleybreed</a> that I dug the lamp parts back out and got it sorted by replacing the LEDs with night light bulbs as in the original lamp (a stroke of genius on Wesley’s part) and rewiring everything with lamp cord, WAGOs, and four screw in bulb connectors.</p><p>The main takeaway from this project was a reminder to always research the problem you’re working on thoroughly: odds are someone has solved it before. Additionally, my first attempt at rewiring was a mess because I tried to cram all the WAGO connectors in the stem of the lamp. Taking the time to start over and rewire everything from scratch, this time with the WAGOs in the base, was way better and saved me a lot of time in the long run.</p><p>I was really happy with the result on this and the homebrew “Artemide” lamp looks great and ended up costing £30 including the cost of the lamp. It could have been even cheaper if I was prepared to wait looking for parts.</p><h3>QR Code Quilt</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*9VzJFbcCHLNr7x68.png" /></figure><p>This QR code project was heaps of fun but a real labour of love. All in all it took nearly a month of evenings after work spent cutting squares out of old conference T-Shirts and painstakingly sewing them together to create the finished panel.</p><p>There were a few tense moments of ironing when I finished but it does actually scan (try scanning the top left panel)! I came into this project without the first idea about quilting and working in an unforgiving material (T-shirt cotton stretches and warps). You can see from the work in progress picture (in the bottom right) that I learned a lot as I went with the panel becoming more square and neat as I progressed. Using a zig-zag stitch helped compensate for the stretchiness of the fabric and made for a neater finish.</p><p>If I were to make this quilt panel like this again, I’d take more care to keep my squares regular and, I would unpick the first section before I started using the zig-zag stitch to keep the overall piece more consistent.</p><h3>Plywood Bike Hanger</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*QJEbDyLCm1p7W6ov.png" /></figure><p>Next is yet another example of me going “I can make you that”. I took a nice piece of plywood my partner picked up for me and turned it into a bike hanger based <a href="https://www.etsy.com/uk/listing/1083425878/hand-finished-bike-wall-mount-road?gpla=1&amp;gao=1&amp;utm_source=connexity&amp;utm_medium=cpc&amp;utm_campaign=shopping_uk_google&amp;utm_content=CjwKCAiA86_JBhAIEiwA4i9JuxH6F6Qwk8HbIMHb5WKP6pIXsiNx9ByW8pKzpgsPCJMVXUpzBUemqhoCVXsQAvD_BwE&amp;gad_source=1&amp;gad_campaignid=23110659535&amp;gbraid=0AAAAAD_PDJmPtH5gLJf1IdNHAC7-qCKaq&amp;gclid=CjwKCAiA86_JBhAIEiwA4i9JuxH6F6Qwk8HbIMHb5WKP6pIXsiNx9ByW8pKzpgsPCJMVXUpzBUemqhoCVXsQAvD_BwE">on this one she saw on Etsy</a>. This came together really quickly and was basically free as I already had all the materials I needed (especially if you ignore the cost of the tools acquired in the years prior!)</p><p>I don’t have a CNC so I marked out the sheet with a speed square and pencil and then cut everything out with a circular saw and a jigsaw for the curves. A bit of filing and sanding was needed to refine the shape of the curves and to get the parts to fit together but after that it was as simple as gluing and screwing the parts together and mounting it on the wall.</p><p>The curved cuts in this are free handed and my jigsaw is less than square (it’s often on loan in the tool library and years of use and abuse have left the blade guide wonky) which made for some wonky cuts. If I did this over again I’d like to try using a holesaw to make the internal curves, and maybe a router with a template for the outer edge to keep things more even. Like the other bike hanger I made last year this is still on the wall which I’m counting as a win.</p><p>Despite my hopes of displaying a nice road bike on the living room wall like the yuppie I am, this hanger is relegated to the garage as it sticks out too far into the room for use in the flat, which is a shame as unlike the other hanger I think it looks great.</p><h3>Laptop Sticker Collection App</h3><p>Much like this post I think laptop stickers are a fun way to keep track of what you’ve got up to over the course of a year. My <a href="https://cowsay.io/?ref=hughevans.dev">friend Leo</a> felt the same and built an interactive <a href="https://github.com/notanotherpizza/stickers?ref=hughevans.dev">web app that allows you to annotate your laptop stickers with where you got them</a>. I helped to contribute some features including support for Markdown in annotations.</p><p>This was my first foray into vibe coding (using Cursor) which proved to be a mistake. I don’t have any experience in web dev with Node.js so I wasn’t able to properly review the AI generated code. The lesson here was to learn the basics before you dive into throwing AI generated spaghetti code at the wall.</p><p>You can check out my writeup of the contributions I made and my live version of the app at <a href="https://hughevans.dev/stickers/">https://hughevans.dev/stickers/</a>. Once I actually took the time to do it properly it was a lot of fun contributing to this project and collaborating with Leo on this.</p><h3>Mitre Saw Stand</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jov5J7APxEnnviwA.jpeg" /></figure><p>My workshop out in the garage was yet again overwhelmed with tools and raw materials earlier this year. I decided to extend my homemade storage with a mitre saw table with some built in storage. I built this out of scrap wood and a kitchen door from the IKEA circular economy store. Learning from my previous mistakes, this time I reduced the width for a snugger fit on the drawers, and added a backstop to each runner to stop the drawers falling out the back.</p><p>This was thrown together in a couple of hours and it shows, although this isn’t much of a concern for what is ultimately a functional piece of workshop furniture. I ran into the classic “measure twice cut once” rookie error when cobbling together the topmost drawer. I ended up with a drawer base a few centimeters short which could have been avoided by taking a little more care marking out my materials, though this was easily fixed by tacking on a strip of wood to fill the gap.</p><p>Overall I’m really happy with this unit, especially with how I managed to get the top leveled properly so the mitre saw base lines up with the adjacent units for a really solid work area for cutting larger pieces.</p><h3>May Day Display</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*3YCC9wWjT7MzSpF6.png" /></figure><p>This year my younger sister was the PettsWood May Queen (for readers outside of the UK, the crowning of a <a href="https://en.wikipedia.org/wiki/May_Queen?ref=hughevans.dev">May Queen</a> is a tradition observed in some places in the UK to mark the coming of spring) and I lent a hand by helping her group with their fundraising by setting <a href="https://pettswoodmayqueen.co.uk/?ref=hughevans.dev">up a website</a> and doing some shifts selling cakes and handicrafts on the stall.</p><p>As part of the full scale family mobilisation to support my sister’s May Queen group my mum came up with an idea for a display stand made of chest of drawers and trellis. This was a fairly elaborate scheme involving cutting a chest of drawers in half and flipping one upside down. This was much more fiddly than it first appeared, but I saw the vision and did my best to help her execute it after finding a free chest of drawers on my street.</p><p>I started strong, neatly separating the top third into a separate piece (see bottom right panel above) but got over confident and ended up with a wonkier cut when removing the bottom third. I added battens to stop the two new mini chest of drawers falling apart, and flipped the drawer slides in the bottom part so that the base became a new top.</p><p>Then came the task of sanding and painting. The lesson here was to spend more time researching how to achieve the desired finish (a rustic white) as the final product left a little to be desired. The end result though was two nice little displays for merchandise that served us well for the month of May as my sister’s group went on to raise over £2000 for the local scout troop to buy a new mini-van.</p><h3>Propeller Cycling Caps for London to Brighton</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*44jLvTFCEPqW3v2V.png" /></figure><p>This year I participated in the London to Brighton bike ride with my partner to raise money for MacMillan. As part of my fundraising efforts I made some silly propeller cycling caps by mashing together some cheap cycling caps and propellor caps.</p><p>This project was heaps of fun especially as the propeller actually spins in the wind and whilst the end result was less than polished, the caps (I ended up making three — one each for my partner and I, and another for our friend) looked great on camera. The caps were made by cutting four cycling caps into strips and hand sewing them back together to achieve a striped effect with the propeller part of a propeller cap stitched into the center.</p><p>I used my cap in the video I made for my <a href="https://www.justgiving.com/page/hugh-evans-london-to-brighton?utm_medium=FA&amp;utm_source=CL">Just Giving page</a> which raised £831, although unfortunately I wasn’t able to wear it for the ride itself as it didn’t fit under my helmet.</p><h3>Bonus: London to Brighton</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*aO9ZwXpg-Cy9tadu.png" /></figure><p>I have nothing to offer the world in terms of meaningful insight into cycling long distances. I had fun though and made it home in one piece.</p><h3>Reviving an old kindle</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*PT50X9Fl3fK890tk.png" /></figure><p>Like a lot of people who make things I practice the skill of hoarding bits and pieces for use in future projects, the skill of course being hoarding things that are actually useful and not just junk, a distinction that isn’t always obvious! As a result I have a box of old electronics and on a whim fished out my old Kindle 2 and decided to see if I could get it up and running to use as an E-Ink display.</p><p>I was pleasantly surprised with how easy it was to get it open and to find a replacement battery online, leaving me to wonder at the (very imaginary) <a href="https://en.wikipedia.org/wiki/Enshittification?ref=hughevans.dev">pre-enshittification</a> era of Amazon. After getting it working I found that I was actually enjoying it as an e-reader although I still need to jailbreak it as I don’t trust Amazon to keep the send to Kindle service working.</p><p>I think I’m not alone in finding the simplicity of older electronics appealing, especially in an e-reader where removing any distractions or notifications is great, plus fixing your old tech is super rewarding. I think I’ll be buying and fixing another Kindle soon as my partner has adopted this one!</p><h3>Smart Bird Feeder</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*AvHdM8YPlSzGlQEV.png" /></figure><p>At time of writing, this smart bird feeder project is still ongoing and has, far and away, been my biggest project of the year.</p><p>This project was kicked off by observing a particularly rotund robin that was visiting my bird feeder. I wanted to know if it was actually particularly heavy or just looked that way, so I did the natural thing and started building a custom bird feeder with a digital scale, webcam, and automatic species detection with image classification models.</p><p>This project has been challenging, requiring learning lots of new skills around working with load cells, data analysis, and training and deploying computer vision models. I’ve already got one talk out of it at the <a href="https://www.meetup.com/pydata-london-meetup/events/310885430/?utm_medium=referral&amp;utm_campaign=share-btn_savedevents_share_modal&amp;utm_source=link&amp;utm_version=v2&amp;member_id=394028516">100th PyData London Meetup</a> and I’m looking forward to expanding upon the work I’ve already done next year. I was lucky to have some really good advice from the lovely people from PyData London on how to do some sensible data analysis to find big robins, which was to use a histogram for the weight data, as opposed to the overcomplicated graph database approach I was using before.</p><p>There were some hard lessons to be learned here including that passively cooled Raspberry Pi 3s can, and will, permanently damage themselves from overheating when running computer vision models under the right circumstances (i.e. manually ignoring thermal warnings). I’ve written at length about everything I’ve done towards this project so far on my blog:</p><ul><li><a href="https://hughevans.dev/load-cell-raspberry-pi/">Weighing birds using a load cell and HX711 signal amplifier</a></li><li><a href="https://hughevans.dev/bird-species-image-classifier/">Detecting bird species from images using an image classifier model</a></li><li><a href="https://hughevans.dev/smart-bird-feeder-histogram/">Data from first phase of the project</a></li></ul><p>…and yes at 26g the Robin was <em>huge</em>!</p><h3>Ikea Tool Roll</h3><p>This was another quick and fun sewing project. I really like IKEA hacks as a genre of make. Often materials sourced from IKEA are incorporated into my projects as they’re cheap and readily available, so it felt natural to try my hand at making something from the classic blue bag as many others have done.</p><p>I took inspiration from <a href="https://beckystern.com/2022/04/03/waxed-canvas-tool-roll/?ref=hughevans.dev">this tool roll pattern</a> by <a href="https://beckystern.com/?ref=hughevans.dev">Becky Stern</a> and sketched out my own design on paper. I selected an old yellow curtain (also from IKEA) for lining, broke down the blue bag for material, and started putting things together. There were a few teething problems as a result of my choice of materials. Even with baking paper to prevent sticking, ironing the bag material was a mistake as it shrinks a lot. Without the option to use an iron, getting nice crisp edges with this sandwich of fabric was basically impossible. The thick layers of curtain and plastic also broke several sewing machine needles which made sewing the pocket part fairly painful. Thank you to <a href="https://nearestnabors.com/?ref=hughevans.dev">Rachel Lee Nabors</a> for gently pointing out that this was my issue, and not as I had first thought that I was installing my needles incorrectly.</p><p>I came across some cool techniques during this project though, one being pinning my pattern piece down on top of the work and sewing through it to get straight lines for the pockets which was really effective. Using the bag handle as a tie to close the roll looked great, really selling the IKEA hack aesthetic. The finished product was actually useful on bike rides for storing allen keys and keeping stuff dry on rainy days.</p><p>Overall I was really happy with this project, it’s pretty rough around the edges but has given me plenty of ideas for future custom tool rolls.</p><h3>Stress sensing Stress Ball</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*OcbavlFgUzDQDwVP.png" /></figure><p>In another example of a silly project that I learned a lot from this <a href="https://hughevans.dev/building-a-stress-sensing-stress-ball-with-raspberry-pi-pico/">Stress Sensing Stress Ball</a> was heaps of fun. I added a force sensitive resistor to this Anaconda stress ball (given to me by the awesome <a href="https://dawnwages.info/?ref=hughevans.dev">Dawn Wages</a> at <a href="https://hughevans.dev/py-con-uk-2025/">PyCon UK</a>) and read the value from it with a Raspberry Pi Pico, which I hadn’t used for more than turning an LED on and off before. I also wrote an entire app using <a href="https://beeware.org/?ref=hughevans.dev">Beeware</a> to visualise the data from the FSR.</p><p><a href="https://hachyderm.io/@HughEvansDev/115312372892348384/embed">https://hachyderm.io/@HughEvansDev/115312372892348384/embed</a></p><p>As is often the case with a project like this I encountered a lot of unknown unknowns on the way which were great learning opportunities, like reading data over stdout from a Pico and packaging Android applications. The takeaway here (and arguably from most of my projects) is build stupid things for the fun of it because you’ll probably learn something new in the process.</p><h3>Disk Invaders</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*kBontZ2tjhn7LQJm.png" /></figure><p>I put together a space invaders clone for work based on a Phaser 3 example and got to see it in action in a custom cabinet made by my colleague <a href="https://www.linkedin.com/in/jeffmeisel/?ref=hughevans.dev">Jeff Meisel</a>. You can <a href="https://hughevans.dev/i-built-an-arcade-game-things-went-wrong/">check out the full write up of this project on my blog</a>.</p><p>I really went to school on this one, I’m not much of a game developer and getting this turned around on a deadline was pretty challenging, and the final product was riddled with bugs. Vibe game development with Claude helped me ship mediocre code fast, but I spent weeks paying for it by unpicking the AI slop from my own ropey code to fix bugs and introduce new features. My colleagues in engineering figured out how to cheat the leaderboard. This was difficult to avoid as the example I cloned ran client side, proving that my security knowledge wasn’t up to scratch for developer-proofing a game aimed at developers!</p><p>It was really fun collaborating with my colleagues to design and build this game, and my favourite part was the amazing cabinet Jeff built. Building with other people is super fun and helps get big projects like this done in time. The game got a great response from conference goers and people got pretty competitive playing it. I’m planning to do a second version in the new year using what I learned building the first version to fix the many, many issues with it including adding server side execution.</p><h3>EMF Camp Collage</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*s--_ror3L0b4K_2s.png" /></figure><p>Electronic Magnetic Field, a hacker festival in Hertfordshire here in the UK, comes but once a year every two years. I went to my first EMF last year with my partner and loved it. I’ve been looking forward to the 2026 edition ever since. Two years is a long time to wait though, so to pass the time I made this collage with some of our souvenirs.</p><p>This was my first time using <a href="https://en.wikipedia.org/wiki/Wheatpaste?ref=hughevans.dev">wheatpaste</a> which is a crazy material: I really like the way it helps to consolidate the colour of the disparate paper elements with the slight tint it adds but I definitely should have mixed it a little thicker, as the moisture warped the thin MDF board I was pasting paper onto.</p><p>I threw together a quick “days to go until EMF” app to run on my <a href="https://tildagon.badge.emfcamp.org/?ref=hughevans.dev">Tildagon </a>badge. I added the badge to the collage as well with a right angle USB-C connector and some M3 bolts I had lying around. One lesson from this project was that I need to get better at taking photos of my project, because I truly have no idea how to photograph a bright subject like the badge.</p><h3>Final thoughts on 2025</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*IFGc85lNoERuAZ-v.png" /></figure><p>This has been a good year.</p><p>That’s not to say it’s been without its challenges: I love my job but it is often very mentally exhausting. I’m tired, really tired. Next year after a rest over the holidays I’m going to have to take better care of myself to avoid burnout, but spending my energy this year has bought me some valuable opportunities.</p><p>This year I spent some wonderful time with old friends and new. A common theme in my projects this year is sage advice and helping hands from some of the lovely and knowledgeable people in my life — I’m better for getting to share and learn from these great people (you know who you are!) Time is always at a premium though, and one disadvantage of seeing a lot of people in a year is not having a lot of time to spend with anyone in particular. My family have been very understanding of my long absences whilst traveling for work, and next year I’m keen to make more time for my loved ones. I’ve also been extremely lucky to have some mentoring from Steve Morris and John Sandall this year, which has been a huge boon for my community work. Steve helped me to completely rebrand AI Signals and prioritise on what needed fixing for the group to thrive. John walked me through how PyData London run events which was vital to growing the team at AI Signals this year.</p><p>Sitting in a cinema in Vilnius watching a brilliant talk from <a href="https://www.linkedin.com/in/polyzos/?ref=hughevans.dev">Giannis Polyzos</a> about Fluss I realised one of the reasons things felt like they were coming together for me this year. Aside from being lucky enough to finally have a job that provides me with enough stability to pursue my interests, this role has given me the chance to spend a good chunk of this year going to conferences, listening and learning from the community. I didn’t go to university, and my previous trips to conferences had mostly been focused on volunteering and booth duty, so for the first time in my adult life this year I found myself sitting in on dozens of hours of technical talks and did my best to soak it all in. Data engineering is a broad field and I don’t think I’ll ever feel comfortable calling myself an expert in even a small part of it, but this year things felt like they were clicking more and more.</p><h3>Looking forward to 2026</h3><p>So, what’s in store for 2026?</p><p>I’m set to spend less time travelling for conferences, hopefully spending more time with friends and family as a result, and focus my professional efforts on content creation and UK based community organising. On the subject of community I’m planning to spend more time contributing to the Python community, including taking a bigger role in helping to organise the PyData London 2026 conference, after being inspired this year by attending <a href="https://hughevans.dev/py-con-uk-2025/">PyCon UK</a> and watching the <a href="https://www.youtube.com/watch?v=GfH4QL4VqJ0&amp;ref=hughevans.dev">Python Documentary</a>. The Python community is full of wonderful, creative, and kind people and Python has been a big part of my life since I wrote my first print(“hello world”) back in 2013, I’m looking forward to giving back to the community as best as I can.</p><p>Another area of community work in the new year I want to work on is learning to be a better facilitator and get out of the other organisers’ way at AI Signals, I’ve been running the group for three years now and have a lot of opinions about how to do it, I want to make it easier for other people to shape the community and grow within it. This goes hand in hand with another big goal for me in the new year which is building my skills as a developer advocate, specifically as I work towards a promotion to staff level. Making the jump from senior to staff is all about collaboration, strategy, and leading without being a <strong><em>Leader. </em></strong>I’ve been working as an independent contributor for a while now but I have a lot to learn when it comes to getting the best results collaborating across functions, helping to design and lead on programs, and measuring the impact of my work.</p><p>One thing I’ve done a lot in 2025 is write and edit blogs. I’ve published 18 blogs to my personal website and helped write and edit several blogs for the streaming team at Aiven. In 2026 I plan to keep writing, I feel like the practice has helped me get better and referring back to my previous posts has been really useful. Next year I’d like to focus on continuing to refine my work and finding my voice beyond just documenting my personal projects. It’s one thing to write guides that impart technical information but I’d like to get better at producing engaging, human writing which feels more important than ever in a world increasingly dominated by AI (even if the AI bubble feels primed to pop in 2026).</p><p>One other area I’d like to focus on in 2026 is making more contributions to open-source. I’ve made small OSS contributions before but I’d like to play a more active role in the projects I work on rather than just writing about them, including reviving my PR to add documentation for the Apache Kafka CLI tools. Working on open-source contributions can be hard and it can be slow with waiting for feedback and making changes, but I’d like to work on a couple of more substantial contributions this year. I’m looking forward to learning more about the tools I work with every day by taking them apart to see what makes them tick.</p><p>That’s a wrap on my last blog for 2025, happy holidays and see you again in 2026!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*A4DcBykM3Ra_t42B7-rkjA.gif" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4357ee11aaba" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I discovered pigeons sabotaging my project with Aiven Free-Tier Kafka]]></title>
            <link>https://medium.com/@hugh.evans.dev/how-i-discovered-pigeons-sabotaging-my-project-with-aiven-free-tier-kafka-2085a7faedd4?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/2085a7faedd4</guid>
            <category><![CDATA[how-to]]></category>
            <category><![CDATA[apache]]></category>
            <category><![CDATA[birding]]></category>
            <category><![CDATA[kafka]]></category>
            <category><![CDATA[free-tier]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Wed, 10 Dec 2025 16:51:47 GMT</pubDate>
            <atom:updated>2025-12-10T17:10:05.531Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JSS8BoYmgcSAdgFnSt7ovg.png" /></figure><h4>How I used Aiven’s free Kafka tier to monitor my smart bird feeder in real-time</h4><h4>TLDR</h4><p>Aiven’s free-tier Kafka offers 5 topics with 1 partition each, 250 KiB/s throughput (in and out), 3-day retention on a 2-node cluster, and hibernates after 5 days of inactivity. This isn’t a serious project, but <strong>can</strong> Kafka still be a useful tool? Absolutely.</p><p><em>For the full, free and up to date version of this blog and more like it check out the </em><a href="https://hughevans.dev/how-i-discovered-pigeons-sabotaging-my-project-with-aiven-free-tier-kafka/"><em>original version on my website</em></a><em>.</em></p><h4>Building a smart bird feeder</h4><p>Over the past few months, my attempt to answer the question ‘is that robin on my bird feeder huge or just fluffy’ sparked a descent into over-engineered bird feeder madness. I did the natural thing and started building a Raspberry Pi powered <a href="https://hughevans.dev/load-cell-raspberry-pi/">custom bird feeder with a digital scale</a>, webcam, and <a href="https://hughevans.dev/bird-species-image-classifier/">automatic species detection with image classification models</a>.</p><p>I succeeded in answering the question, and presented my results at the 100th PyData London Meetup: <a href="https://hughevans.dev/smart-bird-feeder-histogram/">the robin was 26g</a>, that’s a massive robin (150% more than the average <a href="https://www.rspb.org.uk/birds-and-wildlife/robin?ref=hughevans.dev#:~:text=Weight-,14%2D21g,-Wingspan">according to the RSPB</a>)!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Bd23eMuH4DRW4VWn" /></figure><p>Before I could share the project with friends I had to fix the following issues:</p><ul><li>Scale calibration — the value read by the scale would drift over time and produce inaccurate results</li><li>Uptime — the bird feeder would occasionally crash and go offline without me realising</li><li>False positives — the thresholds for the system to log that it has detected a bird were initially too low so I was getting a lot of empty photos back of “birds” that weighed 4g</li><li>Monitoring — my first prototype would produce an image with data in the filename every time it saw a bird, but that left long periods where I had no diagnostic data to troubleshoot the system</li></ul><p>I wouldn’t discover until later that I had a fifth, feathered problem: pigeons were crash-landing my bird feeder, throwing the delicate calibration out of whack. But more on that later.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*79jCs6h4yIiZjCRR" /></figure><p>I used to be an observability engineer (and also a data engineer), so I had a pretty good idea of how to solve this problem: if I instrumented my bird feeder to produce metrics in something like real-time I could spot issues as soon as they occurred and collect data for analysis to help improve the system.</p><p>There was one major problem with my plan. When I was working on observability before, I was doing it at enterprise scale. This meant replication, vendor walled gardens, disaster recovery, heaps of tooling and interconnected systems, and thousand dollar budgets a month chargeable to the client. I don’t have the pocket change to run production grade systems at home — and I was already spending $5 a month to store my images — so I was keen to find a cheaper solution.</p><p>When Aiven launched free-tier Kafka, I realized it was perfect for my observability problem, and without adding any new costs to my already rapidly growing bill of materials.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*rcxdV8O12jJJkMTx" /></figure><p>To get some visibility into my bird feeder problems I set up a quick and dirty data architecture which transmits metrics from the bird feeder to Grafana (an open-source visualization and dashboarding tool) via Kafka.</p><h3>Why Kafka?</h3><p>The free-tier specs are ideal for my project. While 250 KiB/s throughput might not seem like a lot of bandwidth to work with, if you consider that my bird feeder only produces around 1kb/s of message data it’s actually more than enough: hypothetically I could support up to 250 separate bird feeders! Also, because my data volumes are so small, I can use Kafka as my only data source, which dramatically simplifies my architecture.</p><p>If we look at the types of data I’m sending, 5 topics is also plenty to give me clean separation for weight data, motion data, processed bird visits, and system health (including Kafka message throughput) — with a topic to spare if I add another sensor in later. It’s also easy for me to re-use topics for data from separate bird feeders by filtering on location and username tags.</p><p>Durability is important in observability at enterprise scale but it still matters here: when the Pi crashes (and it does), the last N hours of metrics are still in Kafka for post-mortem analysis.</p><p>Finally, I can iterate on Grafana dashboards without touching the bird feeder code thanks to the decoupling Kafka provides to my architecture. It’s worth noting that managed connectors aren’t available through the Aiven platform for free tier but tools like Grafana that manage their own consumers will still work great.</p><p>There are, of course, some limitations to this approach: consuming data directly from Kafka in Grafana is inefficient as it requires rereading from the earliest offset each time the dashboard is reloaded, which makes it difficult to take advantage of the three day retention period. This can be mitigated by using a metrics database like Thanos to store historical data for more efficient retrieval. But for these low data volumes seen in the debugging process, we can get away without using a metric store for now.</p><h3>Producing data to Kafka</h3><p>Getting my various metric data from my Pi into Kafka was surprisingly straightforward thanks to the Python Quick Connect in the Aiven Kafka service. It produces ready-to-use Python code with all the connection details and SSL certificates configured, so I could focus on instrumenting my application rather than wrestling with configuration.</p><p>I structured my data with multiple bird feeders in mind from the start. Each message includes a userId and location field, which means the same Kafka topics can handle data from multiple bird feeders without mixing them up:</p><pre>def send_weight_data_to_kafka(self, weight, timestamp):<br>    &quot;&quot;&quot;Send weight data to Kafka topic&quot;&quot;&quot;<br>    data = {<br>        &#39;userId&#39;: USER_ID,<br>        &#39;weight&#39;: float(weight) if weight is not None else None,<br>        &#39;timestamp&#39;: timestamp.isoformat(),<br>        &#39;location&#39;: FEEDER_LOCATION if FEEDER_LOCATION else None<br>    }<br>    self.send_to_kafka(&quot;weight&quot;, data)</pre><pre>   def send_to_kafka(self, topic, data_dict):<br>        &quot;&quot;&quot;Unified Kafka send with metrics tracking&quot;&quot;&quot;<br>        message = json.dumps(data_dict)<br>        message_bytes = message.encode(&#39;utf-8&#39;)<br>        <br>        producer.send(topic, message_bytes)<br>	  ...<br>        self.send_metrics()</pre><p>My bird feeder now streams weight readings and motion detection data continuously to separate Kafka topics. This gives me a complete timeline of what’s happening at the feeder, not just snapshots when birds are detected, which made fixing the issues with my bird feeder a breeze.</p><h3>Calibrating my sensors using real data</h3><p>One big challenge in this project was picking a sensible activation threshold for triggering the camera and logging a bird sighting. The two sensors I am working with are a load cell which returns a value in grams, and a camera which I can use to detect motion (by measuring the difference between an image taken a few moments ago and what the camera is currently seeing) in pixels.</p><p>You could come across good values for these by doing some research — for example, the lightest bird in the UK is the Goldcrest, coming in at a teensy 5g. Alternatively, you could use trial and error, which is what I initially did for the motion sensing activation. I later took a better approach: I calculated the total area of the image in pixels of the camera sensor, and then experimented with values corresponding to different portions of the image before settling on 30kpx as a reasonable threshold.</p><p>As an approach this wasn’t horrible, but it was less than scientific. Crucially, it was impossible to know if I was missing birds by making my activation threshold too conservative without watching the bird feeder all day. Adding my little observability stack solved this issue by allowing me to compare the live reading of the sensors to the activation threshold in real time, easily identifying missed birds and visually comparing the performance of different values.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*SbaAyjsWXrjOC1qd" /></figure><p>Another huge challenge for calibration was taring (setting it to zero when a bird left or when it first started up) the scale. It was hard to know how different approaches to taring i.e. taring after the bird left, taring every 60 seconds, performed. This made it hard to ensure the weights measured were accurate. Again, by using this lightweight observability setup with Kafka and Grafana, I could easily view the measured weight in real-time and immediately note the impact of different taring strategies.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*YaXRnLKMWhslHPwp" /></figure><h3>Pigeon problems</h3><p>I could see in the data that sometimes the scale would get stuck reading 20g, and by comparing the data from Grafana to my images in Cloudflare I could immediately spot the culprit. Pigeons were landing on the roof of my bird feeder, gradually unsticking it from my window, and upsetting the calibration of the load cell.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_ZlqkqTKfuiOAoSv" /></figure><p>The more I looked into the data the more I could see the impact of my avian adversaries. Most of the birds visiting the feeder would fly in, grab a quick mouthful, and fly away. The pigeons, by contrast, would make themselves at home for as long as it took to eat every scrap of bird seed in the feeder. This was annoying enough when it came to having to frequently refill the feeder, but worse was that the pigeons would hang around through tare cycles causing my data to be polluted with hundreds of pictures of pigeons with inaccurate weights.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bXDkxPXTgEBpxxEt" /></figure><p>I may be too close to this but look at the pigeon’s eyes in the last picture. They know they’re ruining my week.</p><h3>Conclusion</h3><p>Aiven free tier Kafka provides me with everything I need for my personal projects. I don’t need five-nines uptime for bird watching; I need real-time visibility into why my scale keeps drifting. Aiven’s Apache Kafka gives me that for free.</p><p>This isn’t production infrastructure — it’s a 2-node cluster limited to 250KB/s of throughput but, if like me you’re working on hobby projects, it may well be exactly what you need.</p><p>Next up is pigeon proofing my setup: either by modifying my bird feeder or taking the more pigeon friendly approach of remounting the load cell such that pigeon landings stop messing up my readings. Then again maybe more drastic action is needed, look at the malice in those beady eyes. Until then my bird data is full of greedy pigeons, for now.</p><p>What would you build with free Kafka? We’re running a competition with a $5,000 prize for the best project built with Aiven’s free tier Kafka. Whether it’s monitoring your sourdough starter or tracking your cat’s patrol routes, I’d love to see what absurdly over engineered projects you build. Share your project on social media with #FreeTierKafka and tag @Aiven to enter (competition runs until January 31st, 2026).</p><p><a href="https://aiven.io/free-tier-apache-kafka-competition-terms?ref=hughevans.dev">Check out the competition details</a> and <a href="https://console.aiven.io/signup?campaign=kafka_free_tier&amp;ref=hughevans.dev">Aiven free tier Kafka here.</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2085a7faedd4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Migrating from Jekyll to Ghost Blog]]></title>
            <link>https://medium.com/@hugh.evans.dev/migrating-from-jekyll-to-ghost-blog-3a5b30a6afde?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/3a5b30a6afde</guid>
            <category><![CDATA[blogging]]></category>
            <category><![CDATA[how-to]]></category>
            <category><![CDATA[ghost-blog]]></category>
            <category><![CDATA[jekyll]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Sun, 30 Nov 2025 14:10:59 GMT</pubDate>
            <atom:updated>2025-11-30T14:10:59.537Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RKZnZUdHRRStFScZSNbj1Q.png" /></figure><blockquote>For the full, free and up to date version of this blog and more like it check out the original version <a href="https://hughevans.dev/migrating-from-jekyll-to-ghost-blog/">on my website</a>.</blockquote><p>Since 2022, I’ve been hosting my own blog for fun (and zero profit) from a Raspberry Pi 3 in my Homelab. When I first set it up, I was just getting into writing technical content but, now that I write professionally, I found my blog wanting. I needed somewhere I could both publish write-ups of projects I’ve written in my own time but also keep an archive of my past work; including articles, talks, and demos so that they can follow me between jobs. I decided to migrate my simple static Jekyll site to Ghost Blog: a powerful open-source blog and newsletter platform.</p><p>This post is no substitute for the official docs for the Ghost Blog project which you can find <a href="https://docs.ghost.org/?ref=hughevans.dev">here</a> but is intended to share my experiences making the switch and share some of the things I used in case they might help other people with this migration.</p><h3>Why Migrate?</h3><p>My previous blog infrastructure was pretty simple: a static site rendered from Markdown in a GitHub repo using Jekyll with a GitHub action to copy new blogs merged to main over onto my Pi. I also needed some more actions for automatically reposting new blogs to my social pages (<a href="https://hughevans.dev/adventures-in-posse/">which turned out to be surprisingly difficult</a>). This was fine back when I was just getting started and wanted somewhere to keep some of the blogs I’d written for my company at the time from Medium on my own site but wasn’t cutting it for me anymore. Writing blogs directly in Markdown is clunky and converting back and forth from Markdown always seems to require some manual correction.</p><p>I found myself craving the simplicity of a <a href="https://en.wikipedia.org/wiki/WYSIWYG?ref=hughevans.dev">WYSIWYG (What You See Is What You Get)</a> editor that I could easily copy and paste into from docs (if you aren’t drafting your blog as a doc you’re probably missing out on reviews from friends and peers by making it needlessly difficult to leave comments on your work). I also wanted the easy of using pre-built integrations for platforms like Mastodon, Bluesky, Medium, LinkedIn, and easy integration with email tools for newsletters to allow me to employ POSSE (a publishing strategy of posting first to your own site and then syndicating your posts across other platforms) without having to manually maintain integrations.</p><h3>Deploy Ghost</h3><p>I deployed Ghost to my homelab (a cluster of Raspberry Pi 4s) with Docker Compose <a href="https://github.com/robincher/docker-compose-ghost-quickstart/blob/master/docker-compose.yml?ref=hughevans.dev">using the example config from the Ghost documentation</a>. I had to shuffle the newly created Ghost db onto a separate Pi to the one hosting my main nginx server as the read and writes to the db were more than the SD card could handle. This could easily be avoided by picking a USB drive with decent read and write speeds and mounting the docker compose volume there instead but I’m cheap and lazy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_CfCYuhIIhmCMioA.gif" /></figure><p>This creates a Postgres service Ghost uses to store posts and other data from your site. I also configured the newsletter feature in Ghost to use my personal email by default but later signed up for <a href="https://www.mailgun.com/?ref=hughevans.dev">Mailgun</a> to support emailing more users.</p><h3>Migrate Content</h3><h3>Convert posts</h3><p>Migrating content from Jekyll to Ghost is a bit of a pain, a few people have put together tools and made them available for others to use but I couldn’t find anything that worked for my posts consistently so I wrote my own script to convert posts, <a href="https://raw.githubusercontent.com/hevansDev/portfolio-website/refs/heads/ghost-migration/ghost-output/create_ghost_zip.py?ref=hughevans.dev">you can find it here</a>. You can pass the script the path to your Jekyll posts and it will return a JSON file of Ghost posts you can easily import.</p><pre>create_ghost_zip.py ~/blog/_posts ghost-import.json</pre><p>The script handled all 175 of my posts in one go, automatically creating the tag relationships and preserving all my metadata. I found Ghost’s official migration tools helpful for reference, but needed this custom script since my Jekyll setup had some frontmatter fields that required special handling.</p><h3>Copy images to Ghost</h3><p>My script automatically updated my image paths to the correct format for Ghost, so migrating images over was as simple as copying the images directory from my Jekyll repo over into the correct location in my Ghost volume with the following:</p><pre>cp -r ./assets/images/* /path/to/ghost-data/images/</pre><h3>Setting Up Ghost ActivityPub</h3><p>As I mentioned previously, better BlueSky and Mastodon integration was half the reason I was migrating in the first place so it was unfortunate I had so much trouble with this step. <a href="https://activitypub.rocks/?ref=hughevans.dev">ActivityPub</a> is the decentralized protocol that <a href="https://github.com/mastodon/mastodon?ref=hughevans.dev">Mastodon</a> is built on top of so to automatically repost my blogs there I needed to get the Ghost ActivityPub feature working. My struggles here were entirely my own fault for not reading the docs closely enough (<a href="https://forum.ghost.org/t/activitypub-site-missing-error-need-manual-registration-cleanup-for-hughevans-dev/60543/14?ref=hughevans.dev">as you can see in my support ticket</a>) but the Ghost team and community were extremely helpful and helped me fix my config issues — for which I’m extremely grateful.</p><p>The Ghost ActivityPub feature proxies certain requests to Ghost’s hosted service at <a href="http://ap.ghost.org/?ref=hughevans.dev">ap.ghost.org</a>. I made a rookie error and passed the wrong value in for the host header which broke my Fediverse integrations entirely leaving the Network tab in Ghost blank. If you run into a similar issue my fix (<a href="https://www.spectralwebservices.com/?ref=hughevans.dev">as provided by Cathy Sarisky</a>) was to correct this and restart both nginx and Ghost which resolved my issue.</p><pre># ActivityPub - proxy to Ghost&#39;s hosted service<br>location ~ ^/.ghost/activitypub/ {<br>    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;<br>    proxy_set_header X-Forwarded-Proto $scheme;<br>    proxy_set_header X-Real-IP $remote_addr;<br>    proxy_set_header Host $http_host; <br>    proxy_ssl_server_name on;<br>    proxy_pass https://ap.ghost.org;<br>}<br>location ~ ^/.well-known/(webfinger|nodeinfo) {<br>    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;<br>    proxy_set_header X-Forwarded-Proto $scheme;<br>    proxy_set_header X-Real-IP $remote_addr;<br>    proxy_set_header Host $http_host;<br>    proxy_ssl_server_name on;<br>    proxy_pass https://ap.ghost.org;<br>}</pre><p>Restart nginx and Ghost:</p><pre>docker restart nginx<br>docker restart ghost</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*MWvMDy9mYthjmSkk.png" /></figure><h3>Setting Up Bluesky Bridge via Bridgy Fed</h3><p>Getting posts to bridge to Bluesky via Bridgy Fed, also initially stumped me. <a href="https://fed.brid.gy/?ref=hughevans.dev">BridgyFed</a> is a service that automatically mirrors your posts over from ActivityPub to ATProto the decentralised protocol that <a href="https://github.com/bluesky-social/social-app?ref=hughevans.dev">BlueSky</a> is built on to save you the trouble of manually reposting things.After following @bsky.brid.gy@bsky.brid.gy from my Ghost blog’s ActivityPub account, the bridged account was created but posts weren’t flowing through.</p><p>After some troubleshooting and <a href="https://github.com/snarfed/bridgy-fed/issues/2218?ref=hughevans.dev#issuecomment-3571954672">filing a GitHub issue</a>, I discovered the fix: unfollow @bsky.brid.gy@bsky.brid.gy, wait a few minutes, then follow it again. This, in combination with Ryan Barre updating DNS on the Bridgy side of things, resolved what turned out to be a duplicate DID issue where posts were being sent to the wrong Bluesky account.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-8YDyqofbkmhYEso.png" /></figure><h3>Final Thoughts</h3><p>Migrating from Jekyll to Ghost was more painful than I had hoped. Setting up the ActivityPub integration was particularly cumbersome, though this could have been avoided with paying better attention to the docs and ultimately the great community around Ghost helped tremendously.</p><p>Some of the integrations I was hoping for, like Ghost to Medium, just aren’t available natively yet but the brilliant editor in Ghost makes it extremely easy to copy paste an entire blog into Medium where with Jekyll I’d be wrangling with Markdown for a while every time to get things looking right. I’ve yet to make much headway with the newsletter feature and would like to have more options for email senders (mailgun seems to mostly end up in spam despite my best efforts) but ultimately this feature was easy enough to set up and seems future proof enough for me to grow into it.</p><p>I kept my previous Jekyll site configuration safe in case I wanted to migrate back but for now I’m very happy with my Ghost blog and have found it makes it much easier to stage, publish, and share content. If you’re still on Jekyll and have the patience (or the money to make the move, Ghost is widely available as a managed service) I’d thoroughly recommend it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3a5b30a6afde" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Big Data Europe 2025 Highlights]]></title>
            <link>https://medium.com/@hugh.evans.dev/big-data-europe-2025-highlights-02145529297d?source=rss-d8aeb74e08e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/02145529297d</guid>
            <category><![CDATA[community]]></category>
            <category><![CDATA[icebergs]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[kafka]]></category>
            <dc:creator><![CDATA[Hugh Evans]]></dc:creator>
            <pubDate>Sun, 23 Nov 2025 17:33:03 GMT</pubDate>
            <atom:updated>2025-11-23T17:33:03.727Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*4RiafeBogH_ppIM-.png" /></figure><p>Big Data Conference Europe is a conference focused on technical discussions in the areas of Big Data, High Load, Data Science, Machine Learning and AI. I was lucky enough to attend and speak at this year’s event in Vilnius, Lithuania.</p><blockquote><em>For the full, free and up to date version of this blog and more like it check out the original version </em><a href="https://hughevans.dev/big-data-europe-2025-highlights/"><em>on my website</em></a><em>.</em></blockquote><h3>Day One Highlights</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*gVr2mPQJqkjwori3.png" /></figure><p><a href="https://www.linkedin.com/in/frankmunz/">Frank Munz ☁️ 🧱</a>’s talk about building real-time aircraft tracking dashboards with <a href="https://www.linkedin.com/company/apachespark/">Apache Spark</a> Data Pipelines and <a href="https://www.linkedin.com/company/databricks/">Databricks</a> was a tough act to follow but it was really interesting and I enjoyed seeing a different approach to the same problem I would address later in my talk.</p><p><a href="https://www.linkedin.com/in/olenakutsenko/">Olena Kutsenko</a> was amazing as always and brought streaming security to life with her comprehensive talk covering everything you need to consider when trying to keep your company out of the news!</p><p>Even though the <a href="https://www.linkedin.com/company/cloudflare/">Cloudflare</a> outage wreaked havoc with my demo, I still enjoyed taking my real-time, real fast talk on it’s final outing and had some great conversations afterwards with the audience about <a href="https://www.linkedin.com/company/apachekafka/">Apache Kafka</a>, <a href="https://www.linkedin.com/company/clickhouseinc/">ClickHouse</a>, and open-source aviation data (I’ll definitely be trying out GNU Reader now!)</p><p>If you want to try out the demo for yourself (minus the Cloudflare induced data losses) you can check out my <a href="https://hughevans.dev/flight-radar-demo/">post about it here</a>.</p><h3>Day Two Highlights</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ZTUIKofDAVatvu9Z.png" /></figure><p>I’ve been curious about <a href="https://www.linkedin.com/company/apachefluss/">Apache Fluss (Incubating)</a> for a while so <a href="https://www.linkedin.com/in/polyzos/">Giannis Polyzos</a>’ talk on the seven deadly sins of streaming (and how Fluss addresses them) was a perfect start to my second day at BDE.</p><p>Next up there was a super interesting panel on AI and data products featuring <a href="https://www.linkedin.com/in/alessandro-pregnolato-5472501/">Alessandro Pregnolato</a>, <a href="https://www.linkedin.com/in/viktor-kessler/">Viktor Kessler</a>, <a href="https://www.linkedin.com/in/emilianomancuso/">Emiliano Mancuso</a>, <a href="https://www.linkedin.com/in/anton-s-borisov/">Anton Borisov</a>, Pavel Filinov, and <a href="https://www.linkedin.com/in/deveshprasad/">Devesh Prasad</a>. There were plenty of great takeaways from the session but I particularly liked Pavel’s reminder on the importance of failing fast, learning, and not being afraid to end a project in a single meeting.</p><p>Natalia Sokolowska’s session about observability for data quality was eye opening — not least because I hadn’t heard the story about Dr Wolf and spinach before! It was really useful to see the tools <a href="https://www.linkedin.com/company/payback-gmbh/">PAYBACK</a> use to check for data quality at every stage of their pipelines and has definitely given me some ideas for how to save myself from myself in my own projects!</p><h3>Day Three Highlights</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Tr95dbt2Aox6Z_aG.png" /></figure><p>I rounded out my time at BDE with a choice selection of talks from the AI track. <a href="https://www.linkedin.com/in/adebolaolomo/">Adebola Olomo</a>’s talk has both practical advice for preserving your unique voice in a world of AI and the inspiring story of her career that has taken her from learning law in Lagos to founding her own businesses to leading partnerships at <a href="https://www.linkedin.com/company/womeninai/">Women in AI</a>.</p><p>The team at <a href="https://www.linkedin.com/company/netflix/">Netflix</a> have a well deserved reputation for industry leading engineering so it was no surprise that Adrian Taruc’s deep dive into how Netflix built a real-time distributed graph had some great insights into how data engineering gets done at a truly massive scale.</p><p>Finally, <a href="https://www.linkedin.com/in/ana-catarina-de-alencar-6b022990/">Ana Catarina De Alencar</a> gave a brilliant talk about AI that can read emotion and what it means for data privacy and consent. This was a fascinating, thought provoking, and important talk in a world where AI is being integrated into more and more aspects of our lives. If you watch one recording of any of the talks from BDE, watch this one.</p><h3>Bonus:</h3><p><strong>Sightseeing in Vilnius</strong></p><p>I arrived the day before BDE and once I was done preparing for my presentation spent an enjoyable afternoon exploring the city. I have no idea how to be a tourist so I watched <a href="https://www.channel4.com/programmes/travel-man-48-hours-in/on-demand/70460-001">Travel Man: 48 Hours in Vilnius</a> for some inspiration.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*EtXBJyzRRYsXtgue.png" /></figure><p>Cribbing from the Travel Man episode I headed into Vilnius proper for a wander including taking in the views from Gedimas Castle and grabbing a drink in Užupis.<br>After a taste of sight-seeing it was time to head over to <a href="https://techzity.com/">Tech Zity</a> for the <a href="https://luma.com/vakamo">Apache Iceberg Europe Meetup</a>.</p><p><strong>The Apache Iceberg™ Europe Community Meetup</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*O87s2xTDL3I3gq0R.png" /></figure><p><a href="https://www.linkedin.com/in/viktor-kessler/">Viktor Kessler</a> kicked things off with his talk taking us on a journey through the history of data analytics and shared some insights on the future of <a href="https://www.linkedin.com/company/apacheiceberg/">Apache Iceberg</a> REST. The next talk from <a href="https://www.linkedin.com/in/martynas-cibulskis/">Martynas Cibulskis</a> was an impressive showcase of the work the team at <a href="https://www.linkedin.com/company/vinted/">Vinted</a> have done extending the <a href="https://www.linkedin.com/company/apacheiceberg/">Apache Iceberg</a> FileIO abstraction layer.</p><p>We heard about OSS Lakehouse support at <a href="https://www.linkedin.com/company/google-cloud/">Google Cloud</a> from <a href="https://www.linkedin.com/in/kestutisdaugela/">Kęstutis Daugėla</a>, this is especially exciting given that <a href="https://www.linkedin.com/company/google/">Google</a> are now storing over an Exabyte of Iceberg data in GCP! Following this up was a whistle stop tour of the challenges of Iceberg CDC from <a href="https://www.linkedin.com/in/anton-s-borisov/">Anton Borisov</a> who shared the work he’s been doing on this problem at <a href="https://www.linkedin.com/company/heyfresha/">Fresha</a> as well as some hope that there are potentially some improved tools for Iceberg CDC in Iceberg v4.</p><p>Finally CEO at <a href="https://www.linkedin.com/company/lakeops/">LakeOps</a> <a href="https://www.linkedin.com/in/amitgilad/">Amit Gilad</a> dived into the optimisations that are possible when you build observability on your Iceberg metadata. Thanks to <a href="https://www.linkedin.com/in/viktor-kessler/">Viktor Kessler</a> for organising and to <a href="https://www.linkedin.com/company/vakamo-com/">Vakamo</a>, <a href="https://www.linkedin.com/company/confluent/">Confluent</a>, and <a href="https://www.linkedin.com/company/google/">Google</a> for making this event possible. Looking forward to more Iceberg Meetups in Europe next year, perhaps in some warmer weather!</p><h3>Final thoughts</h3><p>I’m extremely grateful to <a href="https://www.linkedin.com/in/au%C5%A1ra-stu%C4%8Dinsk%C4%97-70bbb322b/">Aušra Stučinskė</a> and the rest of the BDE team for organising. It was a fantastic event! And a big thank you to <a href="https://www.linkedin.com/in/francescotisiot/">Francesco Tisiot</a> for suggesting I apply to speak.</p><p>That’s a wrap on BDE and my conference travel for 2025! Looking forward to going home to get some rest and get cracking on putting together some awesome <a href="https://www.linkedin.com/company/apachekafka/">Apache Kafka</a> content with some inspiration from the cool people I’ve got to meet and see speak over the last few months.</p><p>Vilnius is a beautiful city and I hope to come back again and see it in future. Not least because, whilst I did manage to try Šakotis here, I didn’t get time to try Cepelinai.</p><p>Looking forward to next time!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/250/0*JrGf1eWudwyF0xtX.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=02145529297d" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>