<?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[LibertyIT - Medium]]></title>
        <description><![CDATA[Liberty IT thoughts and tech stories on technical leadership, digital transformation and agile software delivery. - Medium]]></description>
        <link>https://medium.com/libertyit?source=rss----a420fb5dad97---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>LibertyIT - Medium</title>
            <link>https://medium.com/libertyit?source=rss----a420fb5dad97---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 03:04:18 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/libertyit" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Tales from Web Summit 2022]]></title>
            <link>https://medium.com/libertyit/tales-from-web-summit-2022-8b545a43834?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/8b545a43834</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[conference]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Andy O'Sullivan]]></dc:creator>
            <pubDate>Mon, 04 Sep 2023 12:36:28 GMT</pubDate>
            <atom:updated>2022-12-12T09:41:08.497Z</atom:updated>
            <content:encoded><![CDATA[<h4>The MetaVerse, AI, Future of Work and more</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_p-0o1xXlNbGkTRvrqijGQ.jpeg" /><figcaption>Like Croke Park on a match day in September</figcaption></figure><p>I recently attended Web Summit, the huge tech conference in Lisbon, Portugal, for the first time since before Covid. With over 70k attendees, 1k speakers, countless startups and more, trying to review everything there would be like trying to review everything on Netflix in one article — impossible — so instead I’ll try give you a flavour of some of the things that caught my attention over the 3 days.</p><h4>The MetaVerse</h4><p>Simply the coolest MetaVerse related item I saw at Web Summit was BMW showcasing its M sports car where the driver wears a VR headset <em>while they are driving the car</em>! Yes, actually driving the car, while the headset shows them a virtual race track in a virtual world. Esports streamer <a href="https://twitter.com/Cailee">Cailee</a> spoke about trying it out: <em>“The most insane experience I’ve ever had”.</em></p><h3>Andy O&#39;Sullivan on Twitter: &quot;&quot;The most insane experience I&#39;ve ever had&quot; @Cailee talking about driving a real BMW M car while wearing a VR headset. It looks insane, but amazing! #WebSummit2022 pic.twitter.com/KM6ImN2UGi / Twitter&quot;</h3><p>&quot;The most insane experience I&#39;ve ever had&quot; @Cailee talking about driving a real BMW M car while wearing a VR headset. It looks insane, but amazing! #WebSummit2022 pic.twitter.com/KM6ImN2UGi</p><p>From everything else MetaVerse related, here’s some interesting nuggets I came across:</p><ul><li>VR is where driverless car companies are chalking up miles as part of their training that wouldn’t be possible in the real world. Here’s an <a href="https://www.technologyreview.com/2022/02/18/1045784/simulation-virtual-world-driverless-car-autonomous-vehicle-school-ai-cruise-waabi/">article</a> I found afterwards to learn more.</li><li>The ‘Industrial MetaVerse’ —e.g. better VR representations of factories to aid in training, safety, onboarding etc. Not a new concept, but easier with the improved VR tech.</li><li>One speaker mentioned a nice approach — learn by doing: scan a factory so it can be represented virtually, then figure out what you can use it for.</li><li>Scanning large amounts of physical locations / objects will lead to ever increasing amounts of data to deal with, leading to the need for better data compression.</li><li>‘Verch’ AKA virtual merch! No need to give people pens and t-shirts anymore!</li></ul><h3>Andy O&#39;Sullivan on Twitter: &quot;Just heard about &quot;verch&quot; - virtual merch, from @SashaWallinger at #WebSummit not one I&#39;d heard before! pic.twitter.com/RuyMRucPlA / Twitter&quot;</h3><p>Just heard about &quot;verch&quot; - virtual merch, from @SashaWallinger at #WebSummit not one I&#39;d heard before! pic.twitter.com/RuyMRucPlA</p><ul><li>A quote I liked “who better to design the MetaVerse than architects” as in actual building architects (not high-brow software engineers!).</li><li>MetaVerse via web-browsers will be important to its adoption i.e. if folks can access virtual content via phones or laptops and not just VR headsets it will spread much quicker.</li><li>Plenty of mentions of Fortnite, Roblox and Minecraft basically being MetaVerses already — millions of people playing, creating together etc.</li><li><a href="https://www.linkedin.com/in/mariehuwe/">Marie Huwe</a> (Dolby) pointed out “When all the tools are different it makes it hard” — i.e. there’s no one, or even two main platforms or tools to build for the MetaVerse.</li><li><a href="https://www.linkedin.com/in/toddngreene/">Todd Greene</a> (CEO, PubNub) with some words of wisdom “there’s going to be lots of apps that suck … apps that don’t suck will emerge”.</li></ul><p>There were a lot of MetaVerse startups, and a few speakers, who were promoting crypto-related MetaVerse products or concepts. I regard nearly all of these are fools’ gold so won’t waste anyone’s time talking about them! Check out my Objective Guide to the MetaVerse though here:</p><p><a href="https://lemarquisofandy.medium.com/the-objective-guide-to-the-metaverse-4db0ff2c364f">The Objective Guide to the MetaVerse</a></p><h4>Artificial Intelligence</h4><p>Pretty much every new project I encounter in work involves some level of AI / Machine Learning so it was no surprise of course to see lots of AI content.</p><p>One interview that stood out was with <a href="https://twitter.com/ciaran_omara">Ciarán O’Mara</a> (Protex AI) who gave a refreshingly honest talk about using AI in enterprises, with plenty of pragmatic points:</p><ul><li>AI is the technology behind the product, but the focus needs to be on the user.</li><li>Trust is key — business partners need to understand that 100% accuracy isn’t possible, and the real world changes a lot continuously e.g. a computer vision system may need to cater for daytime, nighttime, different seasons of the year etc.</li><li>Understand what the real problem is — there could be non-AI solutions.</li><li>Production AI is very different to research AI.</li><li>AI is difficult to explain — “explaining an AWS backend is easier”</li><li>Is the business ready for AI — are there resources available in the process for validation?</li><li>“How to legislate something there’s very hard to understand?”</li></ul><p>All in all, just great to hear someone whose business is AI talking about all the caveats to its usage.</p><h3>Andy O&#39;Sullivan on Twitter: &quot;&quot;Have to set expectations about AI-powered solutions&quot; some refreshing words of wisdom from @ciaran_omara #WebSummit22 pic.twitter.com/a2SrbRpt7y / Twitter&quot;</h3><p>&quot;Have to set expectations about AI-powered solutions&quot; some refreshing words of wisdom from @ciaran_omara #WebSummit22 pic.twitter.com/a2SrbRpt7y</p><h4>Future of Work</h4><p>I managed to fit in a few talks about the future of work, a topic so many of us are pondering currently. TL;DR — nobody really knows what’s going to happen or what’s best — remote, office, hybrid etc.</p><p>Some points I liked however:</p><ul><li>“The most important skill is knowing how to get things done in your company” — from <a href="https://www.linkedin.com/in/dshapero/">Dan Shapero</a> (LinkedIn COO) — making the point that keeping people in your company who actually know how to get things done is very important. That may sound like just common sense, but as someone who works in a large enterprise, the folks who know things like how to comply with the internal AWS security rules etc are worth their weight in gold.</li></ul><h3>Andy O&#39;Sullivan on Twitter: &quot;&quot;The most important skill is knowing how to get things done in your company&quot; Dan Shapiro, LinkedIn COO on the importance of retaining staff #WebSummit2022 pic.twitter.com/TOhrnSHb03 / Twitter&quot;</h3><p>&quot;The most important skill is knowing how to get things done in your company&quot; Dan Shapiro, LinkedIn COO on the importance of retaining staff #WebSummit2022 pic.twitter.com/TOhrnSHb03</p><ul><li>“You can work remotely when you know what you’re doing” <a href="https://www.linkedin.com/in/dennis-schmoltzi-b61b21109">Dennis Schmoltzi</a> talking about what sort of work suits or doesn’t suit remote. I liked this one — it’s definitely easier to work on something you know how to do, but what about new starters to a company who need help?</li><li>“Everyone is a leader in a remote environment” — from <a href="https://www.linkedin.com/in/vstock/?originalSubdomain=de">Vanessa Stock</a> (Pitch). I thought this one really interesting — when everyone is working at home, people are essentially <em>their own leaders </em>so need to be trained as such. They need to know how to motivate themselves, schedule their work and more — things which may be done for them by a manager if they were in an office.</li><li>Companies need to be very intentional about communication and documentation — if folks at home need to know something, there better be a good way of telling them.</li></ul><h4>Other notes</h4><p>Web Summit is like a massive festival where you can see and hear lots of different things. Here’s a few other notes on things that caught my eye over the 3 days:</p><ul><li>Driverless cars — folks are a lot less enthusiastic about the potential as they were a few years ago, with some car companies pulling back from previously stated goals.</li><li>A more likely immediate future is certain aspects of driverless technology being used in certain use-cases i.e. a more organic growth of the tech as oppose to a massive big bang changeover.</li><li>One, potentially controversial, opinion was that the safety goals of driverless cars should be more realistic i.e. better than existing cars, not perfect.</li><li>The founder of Ring, <a href="https://twitter.com/jamiesiminoff">Jamie Siminoff</a>, told a great story of how he invented Ring (smart doorbell with camera) for himself to know who was at the door when he was in his garage. It was only when his wife told him that the device made her feel safer at home that he realised he had something he could turn into a business. This is probably a story he’s told before but I loved it!</li><li>He also told of how investors thought it was a bad idea, as the market for doorbells was so small i.e. how many people buy new doorbells each year?</li><li>I saw a good talk on a robot system for warehouses where the robots <em>can climb</em> the shelves to reduce the need for larger warehouses as they can be higher with smaller ground space — resulting in warehouses closer to cities etc. Not sure if the thought of robots that can climb is comforting or not!</li><li>Fidelidade, one of Portugal’s largest insurance companies, talked about their approach to innovation and their home services product, FIXO, where customers can pay a fixed price for tradesmen to do certain tasks.</li><li>The European Space Agency (ESA) stand had info on very small satellites (Cube-sats) being used to help detect wildfire.</li></ul><h4>Final thoughts</h4><p>Web Summit is a great conference — with enough scale and scope to offer enough items of interest to make the trip worthwhile. Some talks are the usual boring conference fodder but there’s plenty of diamonds amongst the rough. And of course Lisbon itself is as beautiful as ever and a joy to visit — thanks to my employer Liberty IT for sending me, and yes, <a href="https://www.liberty-it.ie/careers/opportunities">we’re hiring</a>!</p><p><em>Please let me know what you think below, or you can reach me or you can get me at </em><a href="https://twitter.com/LeMarquisOfAndy"><em>Twitter</em></a><em> or </em><a href="https://www.linkedin.com/in/andy-o-sullivan-9673019/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8b545a43834" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/tales-from-web-summit-2022-8b545a43834">Tales from Web Summit 2022</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Innovation VS Experimentation]]></title>
            <link>https://medium.com/libertyit/innovation-vs-experimentation-6da462028b24?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/6da462028b24</guid>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Andy O'Sullivan]]></dc:creator>
            <pubDate>Mon, 04 Sep 2023 12:34:51 GMT</pubDate>
            <atom:updated>2023-03-27T13:00:41.796Z</atom:updated>
            <content:encoded><![CDATA[<h4>Are they the same thing?</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8WwP4kBpnDAXUB5hpap1Zg.png" /><figcaption>Image by <a href="https://pixabay.com/users/holdendrils-297699/?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=512503">holdendrils</a> from <a href="https://pixabay.com//?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=512503">Pixabay</a></figcaption></figure><p>I was recently recording a podcast with a colleague in <a href="https://www.liberty-it.ie/about-us/overview">work</a> who asked me “What’s the difference between Innovation and Experimentation? Are they the same thing?” which momentarily took me aback but it’s a great question, and an important one, in the context of enterprises and where to allocate resources.</p><p>In short — no, they aren’t the same thing, so let’s dive right in.</p><h4>Innovation</h4><p>I’ve been defining Innovation for the last while as:</p><blockquote>Innovation is Delivering New Value</blockquote><p><strong>Delivering </strong>— actually building and trying it out. It doesn’t need to be a fully spec’d production system, but to get real world feedback you’re going to need to use a real world solution. Ideation by itself isn’t innovation — we can workshop all day long, but coming up with ideas isn’t enough unless we do something about it.</p><p><strong>New </strong>— it’s something new to your business. It doesn’t need to be new to all of your industry, but if you haven’t done it before and it’s going to improve your business, it’s probably innovation.</p><p>If other parts of your company has done it before however, it’s probably not innovation, it’s implementation. That’s not saying that you shouldn’t do it — but you shouldn’t be allocating “innovation” budget to it; that needs to go on new solutions.</p><p><strong>Value </strong>— this should probably go without saying, but it needs to either make large amounts of $$$ or save large amounts of $$$. If it’s something really cool or flashy, that’s great and may be good for branding or recruitment purposes, but if it’s not actually going to make a material difference to anything, it’s not innovation.</p><p>The “large” amounts is important aswell — innovation isn’t about small incremental changes, it’s about making significant improvements to your business.</p><h4>Experimentation</h4><p>So what’s experimentation, and how is it different to innovation? I like to define it as:</p><blockquote>Innovation is Delivering New Value, Experimentation is Finding Out If There Is Any Value</blockquote><p>Let’s say you want to build a new app for your customers to use — up until now they’ve used a website, maybe you’ve a call centre aswell. The app could make this more convenient, drive more sales, offer better service — let’s do it! This is classic innovation, but it’s not experimentation, as if you do everything right and you get good user adoption, the result can be predicted.</p><p>Experimentation is when you think there’s value in something but need to research and experiment to find out.</p><p>Generative AI is a great example — all the current furore and excitement about Large Language Models, ChatGPT etc has led to businesses wanting to try it out. Even though there are lots of articles, consultants, startups etc promising that it’s going to change the world and your business and that you need it immediately, it requires well structured experimentation to actually find out if there is value to be had, and if so offer an opinion on how it can be realised.</p><p>A former colleague, <a href="https://medium.com/u/a1b86cc5f4b7">Gillian Armstrong</a> had a great way of putting it — exploring (experimentation) is turning uncertainty into risk. Risk can be handled — but we need to first understand it, to define it — so when we’re dealing with a technology or concept where we can’t define the risk because it’s uncertain, we need to remove or reduce the uncertainty, through experimentation.</p><h4>Why is it important to define Innovation and Experimentation?</h4><p>It’s important to understand what innovation is and what innovation isn’t, and what experimentation is and isn’t so you make sure you’re investing appropriately.</p><p>For most enterprises, you’re going to have different types of work:</p><ul><li>Keeping the Lights On — maintaining and supporting your core business. This pays the bills!</li><li>New features, products and services — unless you’re in a remarkably stable industry with almost no change, you’re going to need to keep changing and innovating to keep up or stay ahead of the competition.</li><li>Experimentation — you need to understand new technologies and concepts, so you can know if you need to invest more time and money in them. e.g. Is ChatGPT of real value or not?</li></ul><p>If you don’t define work into different categories like this you’ll end up either stagnating, or spending too much resources on new technology that may not help your business.</p><p>An interesting case study is Meta (Facebook’s) current large investment into the MetaVerse and virtual reality — recent <a href="https://www.cnbc.com/2023/02/01/meta-lost-13point7-billion-on-reality-labs-in-2022-after-metaverse-pivot.html#:~:text=Investing%20Club-,Meta%20lost%20%2413.7%20billion%20on%20Reality%20Labs%20in,Zuckerberg&#39;s%20metaverse%20bet%20gets%20pricier&amp;text=Meta&#39;s%20Reality%20Labs%20unit%20recorded,for%202022%20to%20%2413.72%20billion.">reports of $13 billion loss</a> in Meta’s Reality Labs sounds pretty stark. Maybe they view this as an acceptable price for experimentation; time will tell whether it has been a good or bad investment. What else could they have spent the 13 billion on?!</p><p>Likewise, for every amazing Apple product that gets released, there’s likely many others that were experimented with and not released. Not every company has the $$$ Meta and Apple have of course, so planning and understanding is key for your business!</p><p><strong>Summary</strong></p><ul><li>Innovation — delivering new value.</li><li>Experimentation: finding out if there is value.</li></ul><p><em>Let me know what you think below or you can get me on </em><a href="https://www.linkedin.com/in/andy-o-sullivan-9673019/"><em>LinkedIn</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6da462028b24" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/innovation-vs-experimentation-6da462028b24">Innovation VS Experimentation</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[RIP Google Glass]]></title>
            <link>https://medium.com/libertyit/rip-google-glass-985e1011d638?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/985e1011d638</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[augmented-reality]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[tech]]></category>
            <dc:creator><![CDATA[Andy O'Sullivan]]></dc:creator>
            <pubDate>Mon, 04 Sep 2023 12:33:42 GMT</pubDate>
            <atom:updated>2023-03-22T11:38:03.692Z</atom:updated>
            <content:encoded><![CDATA[<h4>Learnings from a Discontinued Device</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Wc-mcCTMscOY9jTXTC1ytA.png" /><figcaption><a href="https://www.google.com/glass/">source</a></figcaption></figure><p>Google <a href="https://www.google.com/glass/start/">announced</a> last week that its Augmented Reality headset, the Google Glass, was no longer available for sale, and from 15th September this year, will no longer be supported.</p><p>Folks who have followed the Glass device will have guessed that it has been circling the drain for a while now; everyone else probably thought it was long since gone anyway. I’ve had a pair for several years — bought for my <a href="https://www.liberty-it.ie/about-us/overview">company</a> just before Covid hit and pushed to the side as other work got prioritised and we were all figuring out how to work at home. In the last few months however I’ve been experimenting with it, and even though it just got its death notice, I think it’s a nice little, flawed, device that can help us prepare for future AR devices.</p><h4>What was bad about it?</h4><ul><li>Display size and quality — the display is projected slightly above your right eye, so you can go about your business and glance up at it when needed. The display area is quite small, and can be difficult to read depending on the lighting conditions.</li><li>No cellular option — you could connect to a phone’s hotspot or other WiFI, but I would have liked it to have had a standalone sim-card option, it would just be easier.</li><li>The method connecting to WiFi was also slightly crazy — you needed to create a QR code on an online site with the details, then scan that with the device.</li><li>Lack of support and development — there’s been basically radio silence from Google on the Glass for several years, with the official code <a href="https://github.com/googlesamples/glass-enterprise-samples">examples</a> on Github also mostly years without any updates.</li><li>Lack of accessory support — we wanted to get a new set of frames, as we just had the default ones, but locating a pair to purchase was proving difficult, especially in Ireland where we’re based.</li><li>It’s reputation — it had become a bit of a joke in tech circles, an example of a failed product; undeservedly I think. Yes, it was a failure, but it had a lot going for it aswell, which leads to …</li></ul><h4>What was good about it?</h4><ul><li>Very easy to develop for — it ran a version of Android, so developing apps for the Glass was more or less the same as developing apps for an Android mobile device.</li><li>The camera and video camera worked well — easy to use and easy to integrate with apps.</li><li>Likewise with the voice recognition — it worked fine as a user control.</li><li>The side trackpad, which allowed touch controls — via tapping or swiping also worked great.</li><li>The device was comfortable to wear — unlike VR devices like the Oculus Quest, the Glass was lightweight and I never minded wearing it for extended periods of time.</li></ul><h4>New user interface paradigm</h4><p>Most importantly however, I liked the Glass as it was a good way to try out a new user interface paradigm that will likely become more widespread if/when a much better device appears on the market (Apple!).</p><p>The concept of consumers (or enterprise employees) wearing a device on their face, that can display or gather information, and that uses controls like voice recognition or gesture recognition, is probably going to be something mainstream at some stage. Getting an early chance to experiment with building apps for such as device will hopefully prepare us better for future devices.</p><p>As an example, we built a small POC to hook up our Glass device to OpenAI — yes, basically a Tech Buzzword Extravaganza!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_W2s-fyLFD9Kb2Ayt7Ug7g.png" /></figure><p>So, the user asks a question, the Glass sends that to a backend app on AWS, which sends it to an OpenAI API and gets the answer routed back to be displayed on the Glass.</p><p>Even now that the Glass has been discontinued, that user interaction — asking a question and getting an answer displayed back, via an AI-driven API, is likely one that will be repeated on future devices.</p><h4>Conclusion</h4><p>Google Glass had a lot of potential — it was a great piece of technology, but for what ever reason (potentially the lack of support &amp; development, or incorrect marketing) it never caught on. There are multiple other similar devices in market currently, but no clear standout or ‘winner’. Let’s all hope for those Apple iGlasses to arrive soon!</p><p><em>If you’ve any thoughts or comments, please let me know below! You can also find me on </em><a href="https://www.linkedin.com/in/andy-o-sullivan-9673019/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=985e1011d638" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/rip-google-glass-985e1011d638">RIP Google Glass</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Thoughts from the 2022 ARVR Innovate Conference]]></title>
            <link>https://medium.com/libertyit/thoughts-from-the-2022-arvr-innovate-979be53d9755?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/979be53d9755</guid>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[virtual-reality]]></category>
            <category><![CDATA[vr]]></category>
            <category><![CDATA[innovation]]></category>
            <dc:creator><![CDATA[Andy O'Sullivan]]></dc:creator>
            <pubDate>Tue, 24 May 2022 13:22:51 GMT</pubDate>
            <atom:updated>2022-05-24T14:18:49.503Z</atom:updated>
            <content:encoded><![CDATA[<h4>Optimistic but Pragmatic</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eS2n-LDqYmlMJz4AQ5KtHg.png" /></figure><p>I recently attended my first in-person conference since before the pandemic — ARVR Innovate 2022 in Dublin, Ireland — and it was wonderful to be actually back in a room, with other people! Hosted in the RDS, which is a large spacious venue, it never felt crowded or uncomfortable and had the obligatory hand sanitisers stations dotted around.</p><p>ARVR Innovate is a conference that is more focused on the business side of augmented &amp; virtual reality, which is important as a lot of new or emerging technology can get caught up in hype and could do with a dose of business reality. A few years ago, this same conference was full of talks promising the incredible value of AR &amp; VR, but in recent years there has been a more realistic and pragmatic set of views, as folks realise that both technologies are still very much ‘emerging’ and in search of truly valuable use-cases.</p><p>Some of my thoughts after attending the day:</p><h4>The MetaVerse is very much in its early stages</h4><p>The MetaVerse is very much hyped currently, largely driven by large investments by tech companies like Meta &amp; Microsoft, but also (in my opinion) by a lot of folks trying to associate crypto-technology with the MetaVerse — selling virtual land via NFTs, talking about how the MetaVerse will be web3 etc, whereas the truth is the MetaVerse is very much, at worst, just an idea, or at best, a collection of unassociated prototypes.</p><p><a href="https://twitter.com/jamescwatson">James Watson</a> of the Glimpse Group gave a great talk on the MetaVerse where he compared it to a set of unconnected planets, which I thought a good analogy —there’s individual ‘planets’ or applications,</p><p>some with great content, but they’re not connected or related in any meaningful way. There is no “one” MetaVerse — rather multiple early versions of what it may finally end up as.</p><h3>Andy O&#39;Sullivan on Twitter: &quot;&quot;Currently it&#39;s like planets that aren&#39;t connected&quot; Great talk on the MetaVerse by @jamescwatson at #arvrinnovate pic.twitter.com/T5sZadtO6r / Twitter&quot;</h3><p>&quot;Currently it&#39;s like planets that aren&#39;t connected&quot; Great talk on the MetaVerse by @jamescwatson at #arvrinnovate pic.twitter.com/T5sZadtO6r</p><h4>Immersive Tech is like Dog Years</h4><p>One great quote I heard from a panel at the event was “Immersive tech is like dog years” — one year equals about 7 years of advancements! That’s a nice way of explaining about how much change this domain is constantly undergoing. Currently the Oculus Quest 2 is state of the art in VR headsets, but that probably won’t last long — with Meta themselves working on ‘Project Cambria’ AKA Oculus Quest 3, along with other tech companies also working on devices.</p><p>At the conference I tried a Pico Neo 3 — a competitor device whose parent company also owns TikTok, and also a couple of early-stage devices, thanks to the <a href="https://twitter.com/BCFE_Official">Ballyfermot College of Further Education</a> stand, none of which were close to being ready for consumers or enterprise, but did show what’s being explored.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/467/1*JV8eJRiHbz7Cynm5yjay9w.png" /><figcaption>painting in the air with an AR device</figcaption></figure><h4>A Virtual Solution doesn’t always need to be actual VR</h4><p>There was a fantastic example of using technology for your customers in a way that actually helps them, as oppose to be being mainly obsessed with the technology itself. <a href="https://twitter.com/ChupiSweetman">Chupi Sweetman</a> of <a href="https://twitter.com/ChupiJewelry">Chupi Jewellery</a> wanted to create a virtual experience for their customers during the pandemic and first explored creating 3D modelled virtual jewels and rings — only to realise that they would never look the same as actual diamonds etc.</p><p>So they went for a different approach where customers use a web-app to take a photo of their hand — onto which they can then place images of rings, with various options.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ba3eFsa6lPzR5DyTd2izAQ.jpeg" /><figcaption><a href="https://try-on.chupi.com/">https://try-on.chupi.com/</a></figcaption></figure><p>Simple and easy to use, it drove much greater engagement than they had expected.</p><h4>VR is a great platform for the arts</h4><p>A great panel at the conference, moderated by <a href="https://twitter.com/vrcamillecom">Camille Donegan</a>, was about the use of VR in the arts — art, music, theatre etc. One panellist spoke of organising an Opera experience in VR during Covid, and being surprised to see many other similar VR experiences. The pandemic seems to have driven creativity in VR, as theatre companies, artists and others sought out new ways to connect with audiences and attempt to generate new or replacement revenue.</p><h4>Almost no mention of NFTs</h4><p>As I discussed above, there’s a lot of NFT / crypto / web3 hype associated with the MetaVerse and VR, but interestingly there was almost no mention of it at the conference. Most speakers talked about real use-cases, how to try to find value using immersive technology, increasing revenue, improving user-engagement, making remote working more enjoyable and efficient, or the current shortcomings and challenges in the domain. Nobody was extolling the virtues of coin-based virtual real estate, or how web3 will change the world!</p><h4>In Summary</h4><p>I enjoyed the conference, not only for the physical-attendance aspect but also the mixture of optimistic and pragmatic content. Virtual Reality &amp; Augmented Reality do hold a lot of potential value, but finding the right use-cases or opportunities are key, with also the important awareness that the technologies are constantly changing and not ready yet for mass adoption.</p><p>If you’ve any thoughts or comments let me know below, you can find me at <a href="https://twitter.com/LeMarquisOfAndy/">Twitter</a> and <a href="https://www.linkedin.com/in/andy-o-sullivan-9673019/">LinkedIn</a>. My company, Liberty IT is also hiring! More details <a href="https://www.liberty-it.ie/careers/opportunities">here</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=979be53d9755" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/thoughts-from-the-2022-arvr-innovate-979be53d9755">Thoughts from the 2022 ARVR Innovate Conference</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Some thoughts on being a plank]]></title>
            <link>https://medium.com/libertyit/some-thoughts-on-being-a-plank-1434aae57a14?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/1434aae57a14</guid>
            <category><![CDATA[organizational-values]]></category>
            <category><![CDATA[organisational-values]]></category>
            <category><![CDATA[organisational-culture]]></category>
            <category><![CDATA[organizational-culture]]></category>
            <category><![CDATA[ship-of-theseus]]></category>
            <dc:creator><![CDATA[Craig Mccarter]]></dc:creator>
            <pubDate>Wed, 15 Sep 2021 07:36:54 GMT</pubDate>
            <atom:updated>2021-09-15T07:36:54.159Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Kc052DGE9KRAlqS71nxbYw.jpeg" /><figcaption><a href="https://pixabay.com/photos/sunset-ship-sails-boat-sea-ocean-675847/">Source</a></figcaption></figure><p><em>The Ship of Theseus is an artefact in a museum. Over time, its planks of wood rot and are replaced with new planks. When no original plank remains, is it still the Ship of Theseus? Secondly, if those removed planks are restored and reassembled, free of the rot, is that the Ship of Theseus? Neither is the true ship. Both are the true ship.</em></p><p>Anyone who has watched the series will likely recognise the above thought experiment from “the field of identity metaphysics” as introduced in the final episode of Disney/Marvel’s <em>WandaVision</em>. Don’t worry, this is where the references to Marvel stop, so I don’t plan any spoilers or need for you to have any further background knowledge of the superhero universe.</p><p>However, the thought experiment presented did strike a chord with me, one that took a few days to put straight in my own head. Consider we belong to a company that is the ship in the experiment. What are the parts that comprise our company? Is it the offices, the desks, the facilities? Our experience through COVID-19 over the last 18 months has clearly indicated that, no, this is not the case — the company exists independent of the physical office space. I believe it is the individuals, the people, that make our company what it is today. Fitting then that we often hear “our people are our greatest asset”. That’s all well and good, but what does this have to do with the Ship of Theseus?</p><p>Consider individuals who leave our company. I’ll refrain to refer to them as rotten planks for sure, but for one reason or another they are no longer part of the ship. Often, they go to become part of the core frame of new ships, and that’s something we should be proud of. As they leave, we replace those planks with new ones. We have a continual intake of new individuals, new planks, who become part of the ship, replacing in some manner those parts that have gone. Therefore, at some point in time, does the company cease to be the same entity and turn into something different? I don’t believe so. I think there’s something that happens, or rather that <em>needs to happen</em>, for these new planks to become a true part of the whole.</p><p>First up, embracing our values. If we can stretch the ship metaphor a little, imagine our ship is painted grey. I like grey, great colour. But a new plank gets added, and it is a different colour — it’s orange. Ugh, orange. Nothing rhymes with orange. Anyone who looks at our ship now can instantly point to that plank and state “that one is not a part of the ship”. It’s distinctly different. I want to be very careful here — diversity is a very important part of our company and our culture and should be celebrated, but I feel we do need to share a common set of values — those guiding principles that direct how we act and behave towards each other. If we do not share a common set of values, it is like our company exhibits multiple personalities. Your experience with one team could be in stark contrast to your experience with another team, to the point where one might think “these teams are so dissimilar, they must belong to different companies”.</p><p>Secondly, embracing our strategic direction. Half the ship cannot sail in one direction, while the other half sails in another. It will rip itself apart. It is the same with our company — we must be moving in a common direction if we want to be effective. We hear many people ask, “what direction is our company going?”. I’m interested to turn the question around — what direction are <em>you</em> going? Your team? Your unit? Since we collectively are the pieces that comprise the company, whatever direction we are going <em>is</em> the direction the company is going. I feel the more important question we should ask is “am I going in the same direction as everyone else?” Luckily, there are certain planks on our ship who serve the purpose of steering us in a direction, acting as our rudder. However, unlike the ship, in our company we need to be listening to those individuals (appropriately called the Steering Team) and making sure that message filters down through all the layers so that each of the individuals who form our company can align on our direction of motion. I encourage you to understand the strategic direction of your company and see how it aligns to your day-to-day activity. If you cannot make that connection, talk to your immediate leadership. Can they clearly show how the activities of your team or unit reflect on the company strategic direction? If not, you may need to question if you are moving as part of the ship or trying to pull against the direction of motion.</p><p>I’ve been part of my particular ship for many years now. It’s been an interesting journey, watching as we adjust our course and continue to venture into new waters. What I find myself increasingly challenged with is that as I progress through the company, there are more people who are looking to me for direction, and I need to be increasingly conscious of my responsibility to them. As a leader, I’m not here just to facilitate day-to-day delivery among our teams, but also to exhibit our values as part of how I work and continually look for the course adjustments our teams need to make to stay true to our strategic direction.</p><p>I’ve enjoyed playing this thought experiment out in my head, and hopefully you find it thought provoking too. It’s also been a fun metaphor to explore — not often you get to call your executive team a bunch of planks! My parting thought: are you part of the ship contributing to the common direction and values, are you an orange plank or a segment trying to resist the flow, or are you simply a passenger being carried onboard without belonging or exerting any influence?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1434aae57a14" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/some-thoughts-on-being-a-plank-1434aae57a14">Some thoughts on being a plank</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Beginner’s Guide To Machine Learning]]></title>
            <link>https://medium.com/libertyit/machine-learning-guide-level-1-c1ace680c936?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/c1ace680c936</guid>
            <category><![CDATA[unsupervised-learning]]></category>
            <category><![CDATA[machine-learning-ai]]></category>
            <category><![CDATA[beginner-machine-learning]]></category>
            <category><![CDATA[supervised-learning]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Darren Broderick (DBro)]]></dc:creator>
            <pubDate>Fri, 27 Aug 2021 21:18:41 GMT</pubDate>
            <atom:updated>2021-09-03T09:29:30.065Z</atom:updated>
            <content:encoded><![CDATA[<h4>Part 1: Types of Machine Learning</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1002/1*lxbgagqyY3manAT9aqmUwg.png" /><figcaption>Source: chatbotslife</figcaption></figure><p>As a beginner in the world of Machine Learning (ML) I want to document the learnings I come across and explain them in a clear manner. Over a series of articles I plan to walkthrough different elements of Machine Learning from a beginner level up to intermediate.</p><p>This short article will look at only 2 things;</p><ol><li>What is Machine learning?</li></ol><p>2. Four Main Training processes (Supervised, Unsupervised, Semi-Supervised &amp; Reinforcement).</p><h3>What is Machine Learning (ML)?</h3><p>ML is the process of solving a problem through the following steps;</p><ol><li>Gathering a dataset (tabular data)</li><li>Building a statistical model based on the dataset</li><li>The inference(conclusion, or use of) that model to solve a practical problem.</li></ol><h3>A Small ML Example</h3><ol><li>Data gathered from a central server for traffic</li><li>A prediction model is built from each car dataset row, input data(location and speed)</li><li>To estimate upcoming congestion areas so that they can be prevented, a “model” is created.</li></ol><blockquote>A model is simply a file that identifies patterns(makes a prediction) based on the input data.</blockquote><blockquote>Machine learning in such scenarios helps to estimate the regions where congestion could be found.</blockquote><blockquote>Daily traffic experiences (data) is used to train a model to essentially see the future.</blockquote><h3>Machine Learning Types</h3><p>Here are the four types of ML and a short description that helps us identify the differences later on.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/728/1*U0sWK4cfCLXjKNe3iqCU8w.png" /><figcaption>Source: Simplilearn</figcaption></figure><p><strong>Supervised<br></strong>Using data that is labelled, when a dataset is in a collection that is labelled, each row of data is called a <em>feature vector</em>.</p><p><strong>Unsupervised<br></strong>Data here is unlabeled, during training the model discovers patterns and new information on its own.</p><p><strong>Semi</strong>-<strong>supervised<br></strong>The data here has only a small amount that is labelled with the rest being largely unlabelled, as a hybrid between the 2 above it is popular for classifying text documents.</p><p><strong>Reinforcement<br></strong>No data is given to the model is this type, training is very different for reinforcement as the process is <strong>random</strong> and the model makes sequences of decisions to try and get the highest return value possible. Very popular in gaming and a stable for AWS DeepRacer.</p><h3>Supervised Learning</h3><p>We mentioned “<strong>feature vector” </strong>for supervised but as an example, just imagine this data as the height, weight and gender of a person. <br>The most important part here is that each datarow has the same labels so they can be differentiated between their <strong>features</strong>.</p><p>An example of this would be spam detection for emails, you have 2 labels here; <em>spam</em> and <em>not_spam</em>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/920/1*8N9sTER2pdIVocniQrGWdA.png" /><figcaption>Source: TowardsDataScience</figcaption></figure><p>So the goal of supervised learning is to use a dataset to produce a model that takes a feature vector as an input and outputs a prediction.</p><h3>Unsupervised Learning</h3><p>The difference to supervise learning is the data is unlabeled, and the goal is to find unknown patterns within your dataset.</p><p>An example of this would be for fraud detection, (or <strong>anomaly detection</strong> as the hierarchical term).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/762/1*tDjX70Per07tIgBpe6V0Rw.png" /><figcaption>Source: Cloudera</figcaption></figure><p>It’s effective at detecting unseen events or rare occurrences by models made through unsupervised using non-anomalous training examples (datarows). i.e how different is <strong>event x </strong>from a “typical” example in the dataset.</p><p>It’s worth showing the differences between these 2 techniques before explaining semi-supervised.</p><h3>Semi-Supervised Learning</h3><p>So for semi the dataset contains both labeled and unlabeled examples. Usually unlabeled with have a much higher percentage over labeled. The outcome of semi is the same as supervised, that is, to use the <strong>labeled</strong> data to produce a model.</p><p><strong><em>Then what’s the point of semi and having more unlabeled data?</em></strong></p><p>You might think that extra unlabeled data will harm your model training but look at it this way; you’re actually adding <strong>more</strong> information about your problem.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/217/1*rxCxis9tCPWYthV4HKCBiw.png" /><figcaption>Source: MLMinds</figcaption></figure><p>Given that the labeled data came from a similar sample-set as the unlabeled then in effect you’ve <strong>improved</strong> the probability distribution for your entire dataset to leverage.</p><p>If the data you are working with is <strong>challenging</strong> to label, consider semi-supervised learning to help ease the process.</p><h3>Reinforcement Learning</h3><p>Reinforcement learning (RL) is <strong>the training of machine learning models to make a sequence of decisions</strong>.</p><p>It is a subfield of ML where the machine or “agent” in this learning context learns in an environment that can ingest <em>state </em>as a vector of features.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*t2eKR5hJ7O-l7PJI.png" /><figcaption>Source: MathWorks</figcaption></figure><p>A policy is a function (similar to models in supervised learning)</p><p>Taking a feature vector of the state supplied by the environment, the agent produces an action to achieve maximum rewards (usually a float value). As the process is sequential, so to is the decision making. The long term goal of the RL lifecycle is to continually optimise the actions taking to achieve better rewards.</p><p>The agent learns to achieve a goal in an uncertain, potentially complex environment. In RL, an artificial intelligence faces a game-like situation.</p><p>The most popular gamified existence of RL is AWS DeepRacer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*GkYyiazwd9464RhZ.png" /><figcaption>Source: AWS</figcaption></figure><p>For DeepRacer its goal is to;</p><ol><li>Navigate the virtual track (environment)</li><li>Gather info like car position, distance from centerline (state)</li><li>Try a random speed and direction (action)</li><li>Logs its result from the policy (reward)</li><li>Repeat process until termination count (episodes)</li></ol><p>Will likely not go back to RL but if you would like to learn more about Reinforcement Learning through AWS DeepRacer then here is a separate article;</p><p><a href="https://medium.com/libertyit/liberty-deepracer-league-and-reinforcement-machine-learning-481c5df15cda">DeepRacer League and Reinforcement Machine Learning</a></p><p>Thank you, and see you in the next article!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c1ace680c936" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/machine-learning-guide-level-1-c1ace680c936">Beginner’s Guide To Machine Learning</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Podcasting to Connect During Covid]]></title>
            <link>https://medium.com/libertyit/podcasting-to-connect-during-covid-ffff67a03e63?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/ffff67a03e63</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[remote-working]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[podcast]]></category>
            <dc:creator><![CDATA[Andy O'Sullivan]]></dc:creator>
            <pubDate>Wed, 30 Jun 2021 23:57:43 GMT</pubDate>
            <atom:updated>2021-07-01T00:08:53.320Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TUb9gTKqzG-Al91uZrQb1A.png" /><figcaption>icons <a href="https://icons8.com/">source</a></figcaption></figure><p>Like many people around the world, Liberty IT staff have been fortunate enough to have been able to work at home during the pandemic, and from my personal point of view, as a company we’ve adapted well to remote work. Great work still gets done, we’ve onboarded a large amount of new teammates who’ve never even seen our offices and we’re still constantly trying to stay ahead in technology, innovation and more.</p><p>I’ve found however that I’ve, of course, met fewer people across the company over the last year; all the in-office serendipitous or planned meetings are long gone. In corporate-speak “I’ve found my internal network not growing as it had been”, what this really means — I’m chatting to less people.</p><p>Enter the “Get to Know LIT” podcast – an idea from a colleague and ran with a few more colleagues, it’s an internal video-podcast series where we chat to people from all across the organisation. Just casual 20 to 30 minute chats about their roles, interesting projects they’re working on, technologies they’re interested in, and usually anything else that pops into our heads.</p><p>Having been asking people to write blog posts internally for a few years, which is always difficult given the time commitment needed to write a decent post, especially a tech tutorial post, I was surprised how easy it’s been to get people to appear on the podcast.</p><p>It’s likely a combination of a few things – it’s just easy, we record the podcast in simple Teams video calls which people are just used to now anyway, and each episode takes less than an hour to record. I also think that people are enjoying talking with other folks that are outside of their immediate team.</p><p>I’ve really enjoyed recording the episodes and usually end up chatting almost as much before and after the actual recording. 12 episodes in, we’ve chatted to data scientists, product designers, new starters, consultants, engineers at various levels and from different teams, about 5G, serverless, the future of coding, mental health during lockdown, virtual reality and much more. As much as I love my actual job, getting to spend an hour every week or so chatting to other folks about all sorts of different things has been something I’ve looked forward to and has helped add some variety to my office-in-an-attic experience!</p><p>Hopefully the listeners feel the same, and it has been able to shine a light on other folks’ work and lives that they may not have seen otherwise. We’re going to keep recording new episodes, and while they’re currently internal only, I predict it’s only a matter of time before we start publishing them externally aswell.</p><p>Of course, the episodes are also great for new starters — to give them insights about the company, so if you really want to hear them … we’re hiring! <a href="https://www.liberty-it.ie/careers/opportunities">https://www.liberty-it.ie/careers/opportunities</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ffff67a03e63" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/podcasting-to-connect-during-covid-ffff67a03e63">Podcasting to Connect During Covid</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why GraphQL? — Implementation of a GraphQL API on the NestJS framework]]></title>
            <link>https://medium.com/libertyit/why-graphql-implementation-of-a-graphql-api-on-the-nestjs-framework-eace5adb4d1e?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/eace5adb4d1e</guid>
            <category><![CDATA[education]]></category>
            <category><![CDATA[tutorial]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[graphql]]></category>
            <dc:creator><![CDATA[Paul Harbinson]]></dc:creator>
            <pubDate>Tue, 16 Mar 2021 14:44:10 GMT</pubDate>
            <atom:updated>2021-03-17T07:43:29.409Z</atom:updated>
            <content:encoded><![CDATA[<h3>Why GraphQL? Implementation of a GraphQL API on the NestJS framework</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Hbyl7XGOd1-o7RBQzyg_CA.png" /></figure><h4>Introduction</h4><p>GraphQL is an alternative to RESTful API’s. So what benefits or functionality does it provide that REST does not? If you consider over-fetching or under-fetching of data an issue then GraphQL could be the solution. A RESTful endpoint may return dozens of data fields of which you only require a few or you may have to call multiple endpoints to get all the data you need, again fetching even more unnecessary data that has to be stripped out on the client side.</p><p>A GraphQL Schema makes it possible to dig deeper into a single endpoint and fetch all of the data and only the data you require in a single http request. It is also easy to implement real-time updates to your data in the form of subscriptions that use websockets under the hood.</p><p>In the interest of transparency, most apps do not sustain the amount of traffic that would warrant using GraphQL solely to reduce the amount of http requests or the size of data packets. However, from a developer’s perspective I prefer the idea of using a query language on the front-end to select the data I want in one request rather than making multiple requests, mapping the data, and then collating it afterwards.</p><h4>Prerequisites</h4><p>The only prerequisites for this tutorial are an IDE, such as Visual Studio Code, and Node.js installed to host the server, which can be downloaded at <a href="https://nodejs.org/en/download">https://nodejs.org/en/download</a>/. All other download instructions are set out in the detailed steps. A little knowledge of NestJS and Typescript would be useful but not completely essential. GraphQL can be implemented in most languages and I only chose NestJS because I’m using it in my next project. There’s actually a little bit more boilerplate code in NestJS than some other implementations I’ve seen such as with AWS AppSync — A full-stack GraphQL backed application can be implemented very quickly using AWS Amplify, AppSync, and DynamoDB.</p><h4>The Anatomy of GraphQL</h4><p>GraphQL is mainly composed of the Schema and its Resolvers. The Schema describes the make-up of the data that can be fetched by the queries. That is, the types of objects, the relationships between types, and the actions that can executed on the types( e.g. like the CRUD actions in REST — GET, POST etc). GraphQL has 3 main actions; Query (like RESTful GET), Mutation (like POST, PUT, DELETE), and Subscription(Websockets) for real-time updates. The Resolvers are the methods that handle the queries and generate the responses.</p><h4>NestJS</h4><p>Nest (NestJS) is a framework for building Node.js server-side applications. It makes use of HTTP Server frameworks like Express providing a level of abstraction away from the Express API’s but still allowing the developer access to those API’s. If you are a front-end Angular developer you will recognise a lot of the patterns and annotations/decorators used in NestJS.</p><h4>Repo</h4><p>The finished code for the following tutorial can be found here: <a href="https://github.com/PaulHarbinson/nestjs-graphql">https://github.com/PaulHarbinson/nestjs-graphql</a></p><h4>Getting Started</h4><p>In the following set-up, to avoid confusion, I am going to assume the new project name is ‘nestjs-graphql-app’. Replace with your chosen app name.</p><pre>npm i -g <a href="http://twitter.com/nestjs/cli">@nestjs/cli</a><br>nest new nestjs-graphql-app<br>cd nestjs-graphql-app<br>yarn add <a href="http://twitter.com/nestjs/graphql">@nestjs/graphql</a> graphql-tools graphql apollo-server-express class-validator uuid<br>yarn start:dev</pre><p>The above commands install the NestJS Command Line Interface, creates a new Nest project, changes directory into that project, adds dependencies using yarn (or npm if you prefer), and starts the development server. Along with the required Nest and GraphQL dependencies I installed an Express server, a class validator for compile-time type checking, and the uuid package for generating unique ID’s to test the API with.</p><h4>Coding the Project</h4><p>Open the project in the IDE:-</p><pre>code . //opens the project in Visual Studio Code</pre><p>You can delete the following files(and references to them in app.module.ts) if you want since we won’t be using them here:-</p><pre>app.controller.ts<br>app.controller.spec.ts<br>app.service.ts</pre><p>NestJS support two approaches to generating a GraphQL Schema. In the code first approach, you use decorators and TypeScript classes to generate the corresponding GraphQL schema. In the schema first approach, a typePaths property indicates where the GraphQLModule should look for GraphQL SDL schema definition files which need to be written manually. I use the code first approach here. Below will be the completed app.module.ts file. It contains a reference to the UsersModule which we haven’t discussed yet. Everything else is generic:-</p><pre>import { Module } from &#39;@nestjs/common&#39;;<br>import { GraphQLModule } from &#39;@nestjs/graphql&#39;;<br>import { UsersModule } from &#39;./users/users.module&#39;; </pre><pre>@Module({  <br> imports: [    <br>  GraphQLModule.forRoot({autoSchemaFile: true}),    <br>  UsersModule,],  <br> controllers: [],  <br> providers: [],<br>})<br>export class AppModule {}</pre><p>For each new entity or ‘type’, create a folder under /src and in it a module file, a service file, and a resolver file. There are command shortcuts to have Nest generate the code and test classes for you and also update the relevant Module file with references to the new Service and Resolver. For example, if you wanted to create a new Schema ‘type’ called Courses:-</p><pre>nest g module courses<br>nest g service courses<br>nest g resolver courses</pre><p>NestJS will prepend ‘Module’ ‘Service’ and ‘Resolver’ to the given name to form the name of the class, e.g. CoursesModule, CoursesService, and CoursesResolver.</p><p>Below are examples of these files for the Users type:-</p><pre>//users.module.ts</pre><pre>import { Module } from &#39;@nestjs/common&#39;;<br>import { UsersResolver } from &#39;./users.resolver&#39;;<br>import { UsersService } from &#39;./users.service&#39;;<br><br>@Module({<br>    providers: [UsersResolver, UsersService]<br>})<br>export class UsersModule {}</pre><pre>//users.service.ts</pre><pre>import { Injectable } from &#39;@nestjs/common&#39;;<br>import { v4 as uuidv4 } from &#39;uuid&#39;;<br>import { GetUserArgs } from &#39;./dto/args/get-user.args&#39;;<br>import { CreateUserInput } from &#39;./dto/input/create-user.input&#39;;<br>import { User } from &#39;./models/user&#39;;<br><br>@Injectable()<br>export class UsersService {<br>    private users: User[] = [];<br><br>    public createUser(createUserData: CreateUserInput): User {<br>        const user: User = {<br>            userId: uuidv4(),<br>            ...createUserData<br>        }<br><br>        this.users.push(user);<br><br>        return user;<br>    }<br><br>    public getUser(getUserArgs: GetUserArgs): User {<br>        return this.users.find(user =&gt; user.userId === getUserArgs.userId);<br>    }<br>}</pre><p>In the above UsersService only the createUser and getUser actions are displayed.</p><p>This is not representative of a real service in that we are only saving new users to an array. However, it is easy to see how we could be saving to and fetching from a MongoDB database here, for example.</p><p>We can also see how a certain amount of boilerplate code comes into play here. In a folder called ‘models’ we have out User class which contains all the fields that go to make up a User, and we also have a ‘dto’ folder which contains the Data Transfer Objects for decoupling the persistence layer and the API layer. This pattern is commonly used in REST API so is not unique to GraphQL.</p><pre>//user.ts</pre><pre>import { Field, Int, ObjectType } from &quot;@nestjs/graphql&quot;;<br><br>@ObjectType()<br>export class User {<br>    @Field()<br>    userId: string;<br><br>    @Field()<br>    email: string;<br><br>    @Field(() =&gt; Int)<br>    age: number;<br><br>    @Field({ nullable: true })<br>    isSubscribed?: boolean;<br>}</pre><pre>//create-user.input.ts</pre><pre>import { Field, InputType } from &#39;@nestjs/graphql&#39;;<br>import { IsEmail, IsNotEmpty } from &#39;class-validator&#39;;<br><br>@InputType()<br>export class CreateUserInput {<br>    @Field()<br>    @IsNotEmpty()<br>    @IsEmail()<br>    email: string;<br><br>    @Field()<br>    @IsNotEmpty()<br>    age: number;<br>}</pre><pre>//get-user.args.ts</pre><pre>import { ArgsType, Field } from &#39;@nestjs/graphql&#39;;<br>import { IsNotEmpty, isNotEmpty } from &#39;class-validator&#39;;<br><br>@ArgsType()<br>export class GetUserArgs {<br>    @Field()<br>    @IsNotEmpty()<br>    userId: string;<br>}</pre><pre>//users.resolver.ts</pre><pre>import { Resolver, Query, Args, Mutation } from &#39;@nestjs/graphql&#39;;<br>import { GetUserArgs } from &#39;./dto/args/get-user.args&#39;;<br>import { CreateUserInput } from &#39;./dto/input/create-user.input&#39;;<br>import { User } from &#39;./models/user&#39;;<br>import { UsersService } from &#39;./users.service&#39;;<br><br>@Resolver(() =&gt; User)<br>export class UsersResolver {<br>    constructor(private readonly usersService: UsersService) {}<br><br>    @Query(() =&gt; User, { name: &#39;user&#39;, nullable: true })<br>    getUser(@Args() getUserArgs: GetUserArgs): User {<br>        return this.usersService.getUser(getUserArgs);<br>    }<br><br>    @Mutation(() =&gt; User)<br>    createUser(@Args(&#39;createUserData&#39;) createUserData: CreateUserInput): User {<br>        return this.usersService.createUser(createUserData);<br>    }<br>}</pre><p>Again, only the getUser and createUser resolvers are displayed here for brevity. We can see here how the resolver is the link between the GraphQL Schema and the service which connects to the database.</p><p>Speaking of the GraphQL Schema, we haven’t seen what it looks like because the ‘code first’ approach in NestJS generates is on the fly using the decorators and TypeScript classes that we created above. If you want to see the GraphQL Schema file then provide a path to it in the autoSchemaFile property in the GraphQLModule options object in the app.module.ts file.</p><pre>//app.module.ts</pre><pre>import { Module } from &#39;<a href="http://twitter.com/nestjs/common">@nestjs/common</a>&#39;;<br>import { GraphQLModule } from &#39;<a href="http://twitter.com/nestjs/graphql">@nestjs/graphql</a>&#39;;<br>import { join } from &#39;path&#39;;<br>import { UsersModule } from &#39;./users/users.module&#39;;</pre><pre><a href="http://twitter.com/Module">@Module</a>({<br>  imports: [<br>    GraphQLModule.forRoot({<br>      autoSchemaFile: join(process.cwd(), &#39;src/schema.gql&#39;),<br>    }),<br>    UsersModule,<br>  ],<br>  controllers: [],<br>  providers: [],<br>})<br>export class AppModule {}</pre><pre>//schema.gql</pre><pre># ------------------------------------------------------<br># THIS FILE WAS AUTOMATICALLY GENERATED (DO NOT MODIFY)<br># ------------------------------------------------------</pre><pre>type User {<br>  userId: String!<br>  email: String!<br>  age: Int!<br>  isSubscribed: Boolean<br>}</pre><pre>type Query {<br>  user(userId: String!): User<br>}</pre><pre>type Mutation {<br>  createUser(createUserData: CreateUserInput!): User!<br>}</pre><pre>input CreateUserInput {<br>  email: String!<br>  age: Float!<br>}</pre><p>As before, I have removed references to ‘updateUser’, deleteUser’ and the query to fetch all users; full code to be found in the GitHub repo. It mostly just contains the types of objects being worked with and their input fields. The question mark, or bang symbol, represents a non-nullable field.</p><p>Every GraphQL service has a query type and may or may not have a mutation type. These types are the same as a regular object type, but they are special because they define the <em>entry point</em> of every GraphQL query. That brings us to how a front-end application interacts with a GraphQL API. NestJS provides a graphical editor tool called Playground that lets you interact with the API. The query language is quite similar to a json object. You state the type of query(e.g. query, mutation, subscription), the name of the query, any arguments required, and then within the second set of curly braces you state which specific fields you want returned i.e. no over-fetching. If there are relationships between types, e.g. a user is enrolled in many courses, then you have nested entry points e.g. getCourses would be nested within getUser i.e no under-fetching. The screenshots below show the query for creating a user and fetching a user. Finally, another benefit of GraphQL is that you can ask a GraphQL schema for information about what queries it supports. So an editor like Playground can provide detailed documentation on the composition of your schema.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5TslfZp4lVP0v0O3eHQOVw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RP4zqrUcli136D2ygLL9hQ.png" /></figure><h4>Conclusion</h4><p>Will GraphQL replace REST? No. Will it gain popularity in apps where the data required is driven by the front-end? Probably. Personally, I have always disliked having to make a REST call to an endpoint that brings back a huge amount of json with dozens of fields when I am only interested in a few of those fields. I look forward to finding out how practical it is for an app that is dependent on multiple other microservices in the cloud, which, in AWS, would involve calling GraphQL from lambda functions.</p><p><em>Any thoughts or comments let me know below, and see current vacancies in Liberty IT </em><a href="https://www.liberty-it.ie/current-vacancies/"><em>here</em></a><em>!</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=eace5adb4d1e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/why-graphql-implementation-of-a-graphql-api-on-the-nestjs-framework-eace5adb4d1e">Why GraphQL? — Implementation of a GraphQL API on the NestJS framework</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building Modern Applications with AWS CDK (Session 1)]]></title>
            <link>https://medium.com/libertyit/building-modern-applications-with-aws-cdk-session-1-ab72020fed0a?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/ab72020fed0a</guid>
            <category><![CDATA[rekognition]]></category>
            <category><![CDATA[aws-cdk]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[lambda]]></category>
            <category><![CDATA[serverless]]></category>
            <dc:creator><![CDATA[Darren Broderick (DBro)]]></dc:creator>
            <pubDate>Thu, 18 Feb 2021 10:56:15 GMT</pubDate>
            <atom:updated>2023-01-06T23:06:34.208Z</atom:updated>
            <content:encoded><![CDATA[<h3>Building Modern Applications with AWS CDK</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X_ulGpqHMBiICRX-aM4mBA.png" /><figcaption><strong>This article is concise material from “AWS Dev Hour”</strong></figcaption></figure><h3>Introduction</h3><p>This write-up is for anyone looking to start learning about AWS Cloud Development Kit (CDK) and how to create components like s3, dynamoDB and lambda, deploy these and use them in an app.</p><p><strong>What we’ll be building with CDK</strong></p><ul><li>An app to let users upload a photo through a react JS UI, stored in s3 (upcoming sessions).</li><li>It will then trigger a lambda upon the s3 action and store it in dynamoDB (which stores metadata and labels, <em>done this session</em>)</li><li>We can then query(scan) the db to check labels for the image have been added after going through rekognition service.</li></ul><p><strong>Background</strong></p><p>AWS host regular sessions on how to use their services— this post is based on a recent one <a href="https://www.twitch.tv/videos/892197005">https://www.twitch.tv/videos/892197005</a>. The video lasts just over an hour, today we’ll go through the main notes, as well as fixes and ways to avoid certain pitfalls I went into and has a separate working GitHub you can clone:</p><p><a href="https://github.com/DarrenBro/modernApp-AWS-CDK">DarrenBro/modernApp-AWS-CDK</a></p><p>There were however some issues I came across not covered by the video that I’ve included fixes for or tasks you should complete on your end.</p><ol><li>Node Version 15+ had issues with deploying lambda assets <a href="https://github.com/aws/aws-cdk/issues/12536">https://github.com/aws/aws-cdk/issues/12536</a><br><strong>Fix</strong> -&gt;Downgrading node and deleting cdk.out/.cache should stop this issue from occurring. <a href="https://stackoverflow.com/questions/8191459/how-do-i-update-node-js">https://stackoverflow.com/questions/8191459/how-do-i-update-node-js</a></li><li>If you continue without completing steps not shown in the video for getting <em>pillow(PIL)</em> into your directory (steps shown below title <strong>Pillow</strong>) you’ll get the following CloudWatch log error;<br><strong>Fix</strong> -&gt; I’ve included the necessary extracted zip file and put it in the project under the folder “reklayer”.</li><li>The video misses out showing the creation of the lambda layer. This allows the lambda to import ‘<strong>PIL</strong>’ from your attached zip file you created.<br><strong>Fix</strong> -&gt; I’ve attached code snippets that need included in <strong>cdkMainStacks.ts. <br></strong>Otherwise you’ll get the below error in your CloudWatch logs.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/744/1*xH-rmwFRG8FYBafI2aA8zg.png" /><figcaption>‘index’ is referring to the main body of the rekognition lambda</figcaption></figure><p>I’ve flooded <strong>cdkMainStacks.ts </strong>with comments better explaining different components. This file is the main area that most of the code will get added.</p><p>For me this has been a great way to get more hands on with CDK with good progression intervals and building it out yourself is the best way to get familiar with any tech, plus finding solutions for these errors has helped commit it more to memory.</p><p><strong>Tech Stack</strong></p><ol><li>aws-cli<br><a href="https://docs.aws.amazon.com/cli/latest/userguide/install-macos.html">https://docs.aws.amazon.com/cli/latest/userguide/install-macos.html</a></li><li>Node.js to run the project (Version 10.3.0–14.15.3)<br><a href="https://nodejs.org/en/download">https://nodejs.org/en/download</a>/</li><li>IDE that can run any of these languages; <br>Typescript, Javascript, Java, C# or Python<br>(I’m using IntelliJ with Typescript)</li><li>AWS CDK toolkit and bootstrap</li></ol><p><strong>All graphics are taken from the AWS run through video.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BNegi0WW23Sz36Q_CGGAhg.png" /></figure><p><strong>Session 1 </strong>will consist of;</p><ol><li>Installing prerequisites</li><li>Explaining what AWS CDK is</li><li>Creating a new project and getting dependencies installed</li><li>Building 3 components -&gt; s3 / lambda &amp; dynamo db</li><li>Copying an image to scan and scanning the table</li></ol><p><strong>What is AWS CDK?</strong></p><ul><li>A framework we can use to declare our AWS resources</li><li>Sits on top of Cloud Formation <br>(CDK produces the CFT Template after synthesise “cdk synth”)</li><li>Simplifies the process of building out these resource templates</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M27Zsej1nMlFRz9096G8LQ.png" /></figure><p>This is what an application looks like in the CDK, starting with the App, and broken into stacks and constructs(which is what will be built later on).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*53aQK6eXIpvvG7Agc3HpTw.png" /></figure><p><strong>Steps to get started;</strong></p><ol><li>“mkdir cdk-project”</li><li>“cdk init” (shows templates available)</li><li>“cdk init app -l=typescript” (Go with typescript)</li><li>Install aws-iam “npm i @aws-cdk/aws-iam”</li></ol><p><strong>Notes</strong></p><ul><li>app initialised in cdk-projects.ts</li><li>cdk.json tells cdk toolkit how to run the project</li></ul><h3>Pillow</h3><p>The AWS Lambda function uses the Pillow library for the generation of thumbnail images. This library needs to be added into our project so that we can allow the CDK to package it and create an AWS Lambda Layer for us. To do this, you can use the following steps.</p><ol><li>Launch an Amazon EC2 Instance (t2-micro) using the Amazon Linux 2 AMI</li><li>SSH into your instance and run the following commands:</li></ol><pre>sudo yum install -y python3-pip python3 python3-setuptools</pre><pre>python3 -m venv my_app/env</pre><pre>source ~/my_app/env/bin/activate</pre><pre>cd my_app/env</pre><pre>pip3 install pillow</pre><pre>cd /home/ec2-user/my_app/env/lib/python3.7/site-packages</pre><pre>mkdir python &amp;&amp; cp -R ./PIL ./python &amp;&amp; cp -R ./Pillow-8.1.0.dist-info ./python &amp;&amp; cp -R ./Pillow.libs ./python &amp;&amp; zip -r pillow.zip ./python</pre><ol><li>Copy the resulting archive ‘pillow.zip’ to your development environment (we used an Amazon S3 bucket for this)</li><li>Extract the archive into the ‘reklayer’ folder in your project directory</li></ol><p>Your project structure should look something like this:</p><pre>project-root/reklayer/python/PIL<br>project-root/reklayer/python/Pillow-8.1.0.dist-info<br>project-root/reklayer/python/Pillow.libs</pre><ol><li>Remove the python.zip file to clean up</li><li>Terminate the Amazon EC2 Instance that you created to build the archive</li></ol><h3>1st component to build is s3</h3><pre>npm i @aws-cdk/aws-s3</pre><p>In cdkMainStack.ts we’ll be adding;</p><pre>import s3 = require(‘@aws-cdk/aws-s3’);</pre><pre>const imageBucket = new s3.Bucket(this, imageBucketName, {<br>  removalPolicy: cdk.RemovalPolicy.<em>DESTROY<br></em>})<br>new cdk.CfnOutput(this, &#39;imageBucket&#39;, {value: imageBucket.bucketName});</pre><p>A quick note on the constructs here.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4AGz6Nq4Y7AQd2So33Y4BA.png" /></figure><p>So for s3 code above it’s the same principle.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*h34WUOLkrPeZUi3_h_NS_g.png" /></figure><h3>2nd component to build is dynamoDB for storing the image labels</h3><pre>npm i @aws-cdk/aws-dynamodb</pre><pre>const imageTable = new dynamodb.Table(this, &#39;ImageLabels&#39;, {<br>  partitionKey: {name: &#39;image&#39;, type: dynamodb.AttributeType.<em>STRING</em>},<br>  removalPolicy: cdk.RemovalPolicy.<em>DESTROY<br></em>});<br>new cdk.CfnOutput(this, &#39;cdkTable&#39;, {value: imageTable.tableName});</pre><h3>3rd component to build is Lambda layer and Lambda</h3><p>Layer needed to import the PIL library when rekognition lambda is executed.</p><pre>const layer = new lambda.LayerVersion(this, &#39;pil&#39;, {<br>  code: lambda.Code.<em>fromAsset</em>(&#39;reklayer&#39;),<br>  compatibleRuntimes: [lambda.Runtime.<em>PYTHON_3_7</em>],<br>  license: &#39;Apache-2.0&#39;,<br>  description: &#39;A layer to enable the PIL library in our Rekognition Lambda&#39;,<br>});</pre><p>Lambda’s job is to pull an image from the s3 bucket, take the image and send it to the rekognition service to do the image detection.</p><pre>npm i @aws-cdk/aws-lambda @aws-cdk/aws-lambda-event-sources</pre><pre>const rekognitionLambdaFunc = new lambda.Function(this, &#39;rekognitionFunction&#39;, {<br>  code: lambda.Code.<em>fromAsset</em>(&#39;rekognitionlambda&#39;),<br>  runtime: lambda.Runtime.<em>PYTHON_3_7</em>,<br>  handler: &#39;index.handler&#39;,<br>  timeout: Duration.<em>seconds</em>(30),<br>  memorySize: 1024,<br>  layers: [layer],<br>  environment: {<br>    &quot;TABLE&quot;: imageTable.tableName,<br>    &quot;BUCKET&quot;: imageBucket.bucketName,<br>    &quot;RESIZEDBUCKET&quot;: resizedBucket.bucketName<br>  }<br>});</pre><h3>Last part to add is Permissions and Roles</h3><p>To trigger Lambda when object(image) is created in S3.</p><pre>rekognitionLambdaFunc.addEventSource(new event_sources.S3EventSource(imageBucket, {events: [s3.EventType.<em>OBJECT_CREATED</em>]}))</pre><p>Permission to read from s3.</p><pre>imageBucket.grantRead(rekognitionLambdaFunc);<br>resizedBucket.grantPut(rekognitionLambdaFunc);</pre><p>Permission to allow the result of rekognition service from the sent image to be stored in dynamoDB.</p><pre>imageTable.grantWriteData(rekognitionLambdaFunc);</pre><p>Permission policy to allow label detection from rekognition across all resources.</p><pre>rekognitionLambdaFunc.addToRolePolicy(new iam.PolicyStatement({<br>  effect: iam.Effect.<em>ALLOW</em>,<br>  // permission policy to allow label detection from rekognition across all resources<br>  actions: [&#39;rekognition:DetectLabels&#39;],<br>  resources: [&#39;*&#39;]<br>}))</pre><p>And make sure you’ve updated your bucket name to the same as the stack. (Would recommend changing prefix to your name, but stack with add a unique suffix ID value in template after synthesising).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/745/1*Qr3A-JEUw2K1u8Bdst4Pvw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/735/1*oJUrFcyluwDDR4w8jhQjNQ.png" /></figure><p>To create your template for your stacks run:</p><pre>cdk synth</pre><pre>A set of files in a folder called &quot;cdk.out&quot; will be produced with assets and your json template file.</pre><pre>You can re-run cdk synth to populate any code updates.</pre><p><strong>synth </strong>will manufacture a stack (our components) down into fully formed json yaml template (CloudFormation template) which will include all the resources we added above.</p><p>When you see the generated template it will already be quite large. Already cdk has saved us a bunch of effort!</p><p>The diagram below helps show what we’ve done in a smaller scale.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tpu33at0Ko0ca4eirfT4ow.png" /></figure><p>To deploy the resources that we just created run the below command.</p><pre>cdk deploy</pre><p>After seeing the below you just want to hit ‘y’</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LL6EIPs6YBglMmJOfTICvQ.png" /></figure><p>This should take around a few minutes to complete.</p><p>Result of “cdk deploy” -&gt; ✅ cdkMainStack</p><p>Output example</p><pre>cdkMainStack.cdkTable = cdkMainStack-ImageLabelsE524135D-104WIEO86Q2JP</pre><pre>cdkMainStack.imageBucket = cdkmainstack-dbrocdkimagebucketb661dc68-uok6ax6q62sh</pre><p>You can also view your CF stacks in AWS console straight away, pretty nice!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ENKRzYyf589ff9fhpgHIjQ.png" /></figure><p>So that’s your s3, lambda and dynamo db all created for you.</p><p>A quick note on the lambda. For the clients below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/669/1*og5xs2nAarGuK44eN72o2g.png" /></figure><p>As this lambda is being invoked we don’t have to re-instantiate these above clients every time the lambda is run, it can stay ‘hot’ and this helps with performance.</p><p>Test the image upload (now that s3 resources have been created using cdk deploy)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WtebyA6CBuhYmQ33y7k6OQ.png" /></figure><p>You’ll need to use your unique bucket name, e.g.</p><pre>aws s3 cp testimage.jpg s3://cdkmainstack-dbrocdkimagebucketb661dc68-uok6ax6q62sh</pre><h3>Logs</h3><p>Check CloudWatch logs for events and for any errors</p><p>In log group “/aws/lambda/cdkMainStack……” -&gt; click latest log stream to see the image being processed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/836/1*U5llWVQjRG16ANyxcvUMYQ.png" /></figure><p><strong>Or</strong> you can check your new dynamoDB table in the console.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fXdLlqtgXlsqF7FTD9oCmg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OVnr7Uqb0GO4mkyFaESASg.jpeg" /><figcaption>These labels are from an image of Stargate’s Atlantis’ wormhole after it was sent through AWS Rekognition</figcaption></figure><p><strong>Or</strong> you can scan the dynamoDB table, using the output name from the cdk synth. e.g.</p><pre>aws dynamodb scan --table-name cdkMainStack-ImageLabelsE524135D-104WIEO86Q2JP<br></pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/536/1*UenWY9KVWkm6PNLYk9vYhw.png" /><figcaption>And you should see a single output of your image in the terminal</figcaption></figure><p>To get rid of all resources and stop any future running costs run:</p><pre>cdk destroy</pre><pre>(which is also why we added to dynamoDB)<br>removalPolicy: cdk.RemovalPolicy.<em>DESTROY<br></em></pre><p>However, even with this policy added to S3 and using cdk destroy the formation will have a status “<strong>DELETE_FAILED</strong>”</p><p>Without the policy added you will see that the CloudFormation stack is deleted successfully but the S3 bucket remains. Why?</p><p>By default, the Construct that comes from the S3 package, has a default prop called removalPolicy: cdk.RemovalPolicy.RETAIN.</p><p>(CloudFormation does not destroy buckets that are not empty).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/887/1*ECB_hMJHhV4jWywfrlWQ0A.png" /></figure><h3>Option 1: Manually clean the bucket contents before destroying the stack</h3><p>You can do this from the AWS S3 user interface or through the command line, using the AWS CLI:</p><pre># Cleanup bucket contents without removing the bucket itself;<br>aws s3 rm s3://bucket-name --recursive</pre><pre># Then run;<br>cdk destroy</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/810/1*LMtsLQkjJIjkmSvZpTWdHQ.png" /></figure><p>Then the cdk destroy will proceed without errors. However, this can quickly become a tedious activity if your stacks contain multiple S3 buckets or you use stacks as a temporary resource so some automation would help (option 2).</p><h3>Option 2: Automatically clear bucket contents and delete the bucket</h3><p>A 3rd party package called <a href="https://www.npmjs.com/package/@mobileposse/auto-delete-bucket">@mobileposse/auto-delete-bucket</a> provides a custom CDK construct that wraps around the standard S3 construct and internally uses the CloudFormation Custom Resources framework, to trigger an automated bucket contents cleanup when a stack destroy is triggered.</p><p>Install the package:</p><pre>npm i @mobileposse/auto-delete-bucket</pre><p>Use the new CDK construct instead of the standard one:</p><pre>import { AutoDeleteBucket } from &#39;@mobileposse/auto-delete-bucket&#39;</pre><pre>const bucket = new AutoDeleteBucket(this, &#39;my-data-bucket&#39;)</pre><h3>Summary</h3><pre><strong># Add components / build / deploy</strong><br>cdk synth<br>cdk deploy<br>(If you makes any changes, you can run another &#39;cdk deploy&#39;)</pre><pre><strong># Take note of outputs (example)</strong><br>cdkMainStack.cdkTable = cdkMainStack-ImageLabels<br>cdkMainStack.imageBucket = cdkmainstack-dbrocdkimagebucket</pre><pre><strong># Copy test image</strong><br>aws s3 cp testimage.jpg s3://{bucketName}</pre><pre><strong># Check table has been populated, if not check logs</strong><br>aws dynamodb scan --table-name {tableName}</pre><pre><strong># Cleanup tasks</strong><br>aws s3 rm s3://{bucketName} --recursive<br>cdk destroy</pre><p>Some helpful reading links;</p><ul><li><a href="https://docs.aws.amazon.com/cdk/latest/guide/getting_started.html">Getting started with the AWS CDK</a></li><li><a href="https://docs.aws.amazon.com/cdk/latest/guide/bootstrapping.html">Bootstrapping</a></li><li><a href="https://aws.amazon.com/rekognition/getting-started/">Amazon Rekognition - get started - AWS</a></li></ul><p>That’s all for this session, take care!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ab72020fed0a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/building-modern-applications-with-aws-cdk-session-1-ab72020fed0a">Building Modern Applications with AWS CDK (Session 1)</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[All the Machine Learning Announcements at AWS re:Invent 2020]]></title>
            <link>https://medium.com/libertyit/all-the-machine-learning-announcements-at-aws-re-invent-2020-ce3b9f917599?source=rss----a420fb5dad97---4</link>
            <guid isPermaLink="false">https://medium.com/p/ce3b9f917599</guid>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[cloud]]></category>
            <category><![CDATA[mls]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Gillian Armstrong]]></dc:creator>
            <pubDate>Wed, 09 Dec 2020 11:03:59 GMT</pubDate>
            <atom:updated>2020-12-21T21:33:55.195Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Feeling a little overwhelmed by all the Machine Learning announcements so far? Here’s your quick catch-up…</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sRof6JZivU9RMNNHI8EM1A.png" /><figcaption>AWS re:Invent Machine Learning Keynote</figcaption></figure><h3>The Themes</h3><ol><li>Anomaly detection is a big theme — lots of new services (Amazon Lookout for <a href="https://aws.amazon.com/lookout-for-metrics/">Metrics</a>, <a href="https://aws.amazon.com/lookout-for-vision/">Vision</a> and <a href="https://aws.amazon.com/lookout-for-equipment/">Equipment</a>, <a href="https://aws.amazon.com/devops-guru/">DevOps Guru</a>, AWS IoT Device Defender <a href="https://docs.aws.amazon.com/iot/latest/developerguide/dd-detect-ml.html">ML Detector</a>, <a href="https://aws.amazon.com/monitron/">Amazon Monitron</a>, <a href="https://aws.amazon.com/aws-cost-management/aws-cost-anomaly-detection/">AWS Cost Anomaly Detection</a>…) and integrations to make this easier across the board in everything from your own data to industrial equipment.</li><li>AWS ML moves further up the stack — previously the top of the stack was the managed AI Services, we’re seeing a level above that now with products that package multiple services (<a href="https://aws.amazon.com/monitron/">Monitron</a>, <a href="https://aws.amazon.com/healthlake/">HealthLake</a>) to make it easier for anyone to use them.</li><li>Responsible AI edges in there with a new bias detection functionality in SageMaker (<a href="https://aws.amazon.com/sagemaker/clarify/">Clarify</a>). Let’s hope this might start to extend the conversation further into ethics and things like carbon-neutral data centres! (edit: at the <a href="https://virtual.awsevents.com/media/t/1_qsv5itu8/186983933">AWS re:Invent Infrastructure Keynote</a> covered a lot of information around efforts in renewable energy and sustainability, which was great to hear).</li></ol><h3>The Actual Announcements</h3><p>There has been a lot announced — in the run-up to re:Invent, at <a href="https://virtual.awsevents.com/media/t/1_2dqukpps/186983933">Andy Jassy’s keynote</a>, at the first <a href="https://virtual.awsevents.com/media/Machine+Learning+Keynote/1_07cg4srl">machine learning focussed keynote</a> with Swami Sivasubramanian, and even just quietly dropped out there. Here they are all in the one place to save you searching!</p><h4>Instances</h4><ul><li><a href="https://aws.amazon.com/machine-learning/trainium/"><strong>AWS Trainium</strong></a><strong> </strong>— high performance machine learning training chip, custom designed by AWS to offer cost effective and high performance for Machine Learning Training. This uses the same <a href="https://aws.amazon.com/machine-learning/neuron/">AWS Neuron SDK</a> that the <a href="https://aws.amazon.com/machine-learning/inferentia/">AWS Inferentia</a> instances announced last year use.</li><li><a href="https://aws.amazon.com/ec2/instance-types/habana-gaudi/"><strong>Amazon EC2 instances powered by Habana Gaudi</strong></a><strong> </strong>— specifically designed for training deep learning models, these can offer up to 40% better price performance for deep learning models.</li><li><strong>Improved Tagging support for </strong><a href="https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/AMIs.html"><strong>Amazon Machine Images</strong></a><strong> and </strong><a href="https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/EBSSnapshots.html"><strong>EBS Snapshots</strong></a><strong> </strong>— new support for tag-on-create and tag-based access control.</li><li><a href="https://pages.awscloud.com/amazon-ec2-p4d.html"><strong>Amazon EC2 P4d</strong></a><strong> </strong>instance support in Amazon SageMaker.</li></ul><h4>Amazon SageMaker</h4><ul><li><a href="https://aws.amazon.com/sagemaker/clarify/"><strong>SageMaker Clarify</strong></a><strong> </strong>— tooling to provide greater visibility into training data and models to identify and limit bias and explain predictions.</li><li><a href="https://aws.amazon.com/sagemaker/data-wrangler/"><strong>SageMaker Data Wrangler </strong></a>— a single visual interface for data preparation workflow, including data selection, cleansing, exploration, and visualisation with over 300 built-in data transformations so you can quickly normalise, transform, and combine features without having to write any code</li><li><a href="https://aws.amazon.com/sagemaker/feature-store/"><strong>SageMaker Feature Store</strong></a> — fully managed, purpose-built repository to store, update, retrieve, and share machine learning features.</li><li><a href="https://aws.amazon.com/sagemaker/pipelines/"><strong>SageMaker Pipelines</strong></a> — continuous integration and continuous delivery (CI/CD) service for machine learning, allowing you to create, automate, and manage end-to-end ML workflows at scale.</li><li><a href="https://aws.amazon.com/sagemaker/distributed-training/"><strong>SageMaker Distributed Training</strong></a> — easily add data parallelism or model parallelism to your PyTorch and TensorFlow training scripts for faster training.</li><li><a href="https://aws.amazon.com/sagemaker/edge-manager/"><strong>SageMaker Edge Manager</strong></a> — model management and monitoring across fleets of edge devices such as smart cameras, robots, personal computers, and mobile devices.</li><li><strong>Deep Profiling added to </strong><a href="https://aws.amazon.com/sagemaker/debugger/"><strong>SageMaker Debugger</strong></a><strong> </strong>— new capability to visually profile and monitor hardware and system resource utilization in during model training.</li><li><strong>Additional </strong><a href="https://docs.aws.amazon.com/sagemaker/latest/dg/model-monitor.html"><strong>SageMaker Model Monitoring </strong></a><strong>capabilities</strong> — new capabilities to detect drift in model quality, model bias, and feature importance in running production models.</li><li><a href="https://aws.amazon.com/sagemaker/getting-started/"><strong>SageMaker JumpStart</strong></a><strong> </strong>— quick-start set of one-click deployable solutions for over 150 pre-trained models.</li><li><a href="https://docs.aws.amazon.com/sagemaker/latest/dg/interface-vpc-endpoint.html"><strong>Secure SageMaker Studio Access</strong></a><strong> </strong>available from <a href="https://aws.amazon.com/vpc/">Amazon Virtual Private Cloud (VPC)</a> through <a href="https://aws.amazon.com/privatelink/">AWS PrivateLink</a>, allowing access without going through the public internet.</li><li><a href="https://aws.amazon.com/sagemaker/autopilot/"><strong>Amazon SageMaker Autopilot</strong></a> adds Deep Learning Models allowing for the automatic creation of accurate models for problems with multi-dimensional, multi-class datasets.</li></ul><h4>Data, Search and Analytics</h4><ul><li><a href="https://aws.amazon.com/redshift/features/redshift-ml/"><strong>Amazon Redshift ML</strong></a><strong> </strong>— create, train, and apply machine learning models using familiar SQL commands in Amazon Redshift data warehouses using SageMaker Autopilot.</li><li><a href="https://aws.amazon.com/neptune/machine-learning/"><strong>Amazon Neptune ML</strong></a><strong> —</strong> run machine learning directly using Neptune APIs and queries using<strong> </strong>Graph Neural Networks (GNNs) purpose-built for graph data.</li><li><a href="https://aws.amazon.com/quicksight/q/"><strong>Amazon Quicksight Q</strong></a><strong> — </strong>automatically understands the relationships across your data using ML, allowing for natural language business questions about your data to be answered instantly.</li><li><strong>New </strong><a href="https://aws.amazon.com/kendra/connectors/"><strong>Connector Librar</strong></a><strong>y for </strong><a href="https://aws.amazon.com/kendra/"><strong>Amazon Kendra</strong></a><strong> </strong>— new native and partner connectors, including Google Drive, can now be indexed and searched through Kendra.</li><li><strong>Incremental Learning for</strong> <a href="https://aws.amazon.com/kendra/"><strong>Amazon Kendra</strong></a><strong> — </strong>automatic<strong> </strong>incremental learning to continuously tune and optimize search results based on end-user search patterns and feedback.</li><li><strong>Custom synonyms in </strong><a href="https://aws.amazon.com/kendra/"><strong>Amazon Kendra</strong></a> – expand the search vocabulary with your own specific business terminology</li><li><a href="https://aws.amazon.com/healthlake/"><strong>Amazon HealthLake</strong></a> <strong>(Preview) — </strong>secure HIPAA-eligible data lake designed for healthcare providers, health insurance companies, and pharmaceutical companies. It automatically uses specialised machine learning models, like natural language processing, to automatically extract meaningful medical information from the data and provides powerful query and search capabilities to remove the heavy lifting of organising, indexing, and structuring patient information to provide a complete view of the health of individual patients and entire patient populations in a secure, compliant, and auditable manner.</li><li><a href="https://aws.amazon.com/lookout-for-metrics/"><strong>Amazon Lookout for Metrics</strong> </a>— automatically detect, prioritise and diagnose anomalies in time series data from various Amazon data stores and third-party SaaS applications. This pairs with <a href="https://aws.amazon.com/appflow/">Amazon AppFlow</a> to get data in a few clicks from those data sources.</li><li><strong>Accuracy measurements for individual items in</strong> <a href="https://aws.amazon.com/forecast/"><strong>Amazon Forecast</strong></a><strong> — </strong>view accuracy for individual items and export forecasts generated during training by being able to access the forecasted values from Forecast’s internal testing of splitting the data into training and <a href="https://docs.aws.amazon.com/forecast/latest/dg/metrics.html">backtest</a> data groups to compare forecasts versus observed data and item-level accuracy metrics.</li><li><strong>Weather Index added to </strong><a href="https://aws.amazon.com/forecast/"><strong>Amazon Forecast</strong></a><strong> — </strong>automatically including the latest local weather information to your demand forecasts with one click and at no extra cost.</li><li><a href="https://aws.amazon.com/emr/features/studio/"><strong>Amazon EMR Studio</strong></a> <strong>(Preview)</strong>— integrated development environment (IDE) running on EMR clusters for data scientists and data engineers, providing fully managed Jupyter Notebooks, and tools like Spark UI and YARN Timeline Service and the ability to execute parameterized notebooks as part of scheduled workflows using orchestration services like Apache Airflow or Amazon Managed Workflows for Apache Airflow.</li></ul><h4>Development/Application Performance</h4><ul><li><a href="https://aws.amazon.com/devops-guru/"><strong>Amazon DevOps Guru</strong></a><strong> (Preview)</strong>— automatic anomaly detection in your application’s operational performance and availability, with recommendations and a visualisation dashboard.</li><li><strong>Memory Profiling in </strong><a href="https://aws.amazon.com/codeguru/"><strong>Amazon CodeGuru</strong></a><strong> — </strong>new capabilities added to profile your application’s memory, giving you a consolidated view of the heap, both as a summary and a time series.</li><li><strong>Python support in</strong> <a href="https://aws.amazon.com/codeguru/"><strong>Amazon CodeGuru</strong></a><strong> — </strong>now Python code as well as Java can be Reviewed and Profiled.</li><li><strong>Security Detectors in</strong> <a href="https://aws.amazon.com/codeguru/"><strong>Amazon CodeGuru</strong></a><strong> —</strong> help identify security risks from the top ten Open Web Application Security Project (OWASP) categories, security best practices for AWS APIs, and common Java crypto libraries.</li><li><strong>Code Quality Detector in</strong> <a href="https://aws.amazon.com/codeguru/"><strong>Amazon CodeGuru</strong></a><strong> — </strong>new metrics to help manage technical debt and codebase maintainability — method source lines of code, cyclomatic complexity, method fan out, class fan out and class cohesion.</li><li><strong>Tag Support in</strong> <a href="https://aws.amazon.com/codeguru/"><strong>Amazon CodeGuru</strong></a></li></ul><h4>Industrial/Manufacturing/Commercial</h4><ul><li><a href="https://aws.amazon.com/lookout-for-vision/"><strong>Amazon Lookout for Vision</strong></a><strong> </strong>— service to find visual defects in industrial products, using computer vision to identify missing components in products, damage to vehicles or structures, irregularities in production lines, and even miniscule defects in silicon wafers — or any other physical item where quality is important.</li><li><a href="https://aws.amazon.com/lookout-for-equipment/"><strong>Amazon Lookout for Equipment</strong></a><strong> (Preview)</strong> — service which provides customers with existing sensors on their industrial equipment, a way to send their sensor data to AWS to build machine learning models for them and return predictions to detect abnormal equipment behavior.</li></ul><h4>Devices</h4><ul><li><a href="https://aws.amazon.com/monitron/"><strong>Amazon Monitron</strong></a><strong> </strong>— end-to-end system that uses ML to detect abnormal behaviour in industrial machinery. It bundles sensors, a gateway device to transfer sensor data to the cloud, the Monitor service itself running on that data, and a companion mobile app to set up the devices and receive reports on operating behaviour and alerts to potential failures in your machinery. It requires no specialist knowledge to set up or use.</li><li><a href="https://aws.amazon.com/panorama/appliance/"><strong>AWS Panorama Appliance</strong></a><strong> — </strong>a machine learning appliance that allows you to to add CV to your existing IP cameras, and automate tasks that traditionally required human inspection and monitoring.</li><li><a href="https://aws.amazon.com/panorama/panorama-device-sdk/"><strong>AWS Panorama SDK</strong></a><strong> </strong>— allows device manufacturers to add AWS Panorama to their own edge devices and smart cameras.</li><li><strong>New divisions, rewards, and community leagues for </strong><a href="https://aws.amazon.com/deepracer/"><strong>AWS DeepRacer </strong></a><strong>League</strong></li><li><strong>New </strong><a href="https://docs.aws.amazon.com/iot/latest/developerguide/dd-detect-ml.html"><strong>ML Detector</strong></a><strong> feature for </strong><a href="https://aws.amazon.com/iot-device-defender/"><strong>AWS IoT Device Defender</strong></a><strong> (Preview)</strong> — monitor your IoT devices by identifying and alarming on anomalous datapoints (e.g., authorization failure counts, message sent counts) using ML models that are automatically retrained on your device data daily.</li><li><a href="https://aws.amazon.com/deepcomposer/"><strong>AWS DeepComposer</strong></a> adds a new learning capsule that dives deep into Transformer models.</li></ul><h4>Amazon Connect</h4><ul><li><a href="https://aws.amazon.com/connect/wisdom/"><strong>Amazon Connect Wisdom</strong></a><strong> (Preview)</strong> — natural language search for agents across connected repositories, and real-time speech analytics to detect customer issues during calls and provide agents recommendations and answers.</li><li><a href="https://aws.amazon.com/connect/customer-profiles/"><strong>Amazon Connect Customer Profiles</strong></a><strong> </strong>— automatically brings together customer information from multiple applications into a unified customer profile</li><li><strong>Real-Time </strong><a href="https://aws.amazon.com/connect/contact-lens/"><strong>Contact Lens For Amazon Connect</strong></a><strong> </strong>— extending the current Contact Lens functionality by adding the ability to get alerted to issues during live customer calls and deliver proactive assistance to agents while calls are in progress.</li><li><a href="https://aws.amazon.com/connect/tasks/"><strong>Amazon Connect Tasks</strong></a> — allows agents to create and complete tasks in the same user interface they take calls and chats.</li><li><a href="https://aws.amazon.com/connect/voice-id/"><strong>Amazon Connect Voice ID </strong></a><strong>(Preview)</strong> — real-time caller authentication using ML-powered voice analysis.</li><li><strong>Added</strong> <strong>inbound telephony</strong> for Argentina, Chile, Mexico, and Peru.</li><li><strong>Reduced telephony rates </strong>in Argentina, Chile, Mexico, and Peru from the US East (N. Virginia) and US West (Oregon) regions.</li><li><strong>Apple Business Chat support (Preview) — </strong>provide convenient and familiar customer service directly through the Apple Messages app</li><li>Support for <a href="https://aws.amazon.com/lex/">Amazon Lex</a> chatbots with <strong>Latin American Spanish and German</strong></li></ul><p><strong>Conversation and Text</strong></p><ul><li><a href="https://aws.amazon.com/lex/"><strong>Amazon Lex</strong></a><strong> additional language support — </strong>German and American Spanish added.</li><li><a href="https://aws.amazon.com/transcribe/"><strong>Amazon Transcribe</strong></a><strong> streaming transcription support extended </strong>— Brazilian Portuguese, Japanese and Korean in the South America (Sao Paulo), Asia Pacific (Tokyo) and Asia Pacific (Seoul) regions now supported.</li><li><a href="https://aws.amazon.com/transcribe/"><strong>Amazon Transcribe</strong></a><strong> adds native support for Ogg opus and FLAC encoded audio</strong></li><li><a href="https://aws.amazon.com/transcribe/medical/"><strong>Amazon Transcribe Medical</strong></a><strong> extended specialities</strong> — new streaming transcription support for additional medical specialties: cardiology, oncology, neurology, radiology, and urology</li><li><a href="https://aws.amazon.com/translate/"><strong>Amazon Translate</strong></a><strong> additional language support</strong> — Armenian, Catalan, Gujarati, Haitian, Icelandic, Kannada, Kazakh, Lithuanian, Malayalam, Macedonian, Maltese, Mongolian, Sinhala, Telugu, Uzbek, and Welsh added.</li></ul><p>Other</p><ul><li><a href="https://aws.amazon.com/aws-cost-management/aws-cost-anomaly-detection/"><strong>AWS Cost Anomaly Detection</strong></a> becomes Generally Available — a free service that using ML to monitor your spending patterns within your AWS Account and detect anomalous spend and provide root cause analysis.</li></ul><p><strong><em>Note</em></strong><em>: This list covers everything from </em><strong><em>Nov 24th to Dec 21st</em></strong><em> – there are of course lots of new features released continuously through the year.</em></p><p>To see some reactions to the Machine Learning keynote you can watch Mike Chamber get together AWS Heroes immediate reactions in this <a href="https://youtu.be/BDe8833La5A">video</a>, read <a href="https://dev.to/aletheia">Luca Bianchi</a>’s <a href="https://dev.to/aws-heroes/aws-heroes-reactions-to-swami-sivasubramanian-keynote-on-machine-learning-cjj">summary</a> or <a href="https://zamirajaupaj.medium.com/">Zamira Jaupaj</a>’s<strong> </strong>live <a href="https://zamirajaupaj.medium.com/machine-learning-keynote-liveblog-re-invent-2020-9abab404a280">blog post</a>.</p><p><strong>I think the overall message is still the same as we have seen for the past few years — AWS continues on its goal of making machine learning as expansive and impactful as it can be by not just creating low-level tooling, but creating a wide array of tools and packages that make it easier for increasingly more and more people to access and use machine learning in their own environments.</strong></p><p><strong>If you are interested in machine learning and AWS, do connect with me on </strong><a href="https://twitter.com/virtualgill"><strong>twitter</strong></a><strong> and let’s have a chat!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ce3b9f917599" width="1" height="1" alt=""><hr><p><a href="https://medium.com/libertyit/all-the-machine-learning-announcements-at-aws-re-invent-2020-ce3b9f917599">All the Machine Learning Announcements at AWS re:Invent 2020</a> was originally published in <a href="https://medium.com/libertyit">LibertyIT</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>