<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Alexandr Wang on Medium]]></title>
        <description><![CDATA[Stories by Alexandr Wang on Medium]]></description>
        <link>https://medium.com/@alexandrwang?source=rss-97d94781bd42------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*tynyPDNtk00c9Pii4muC5w@2x.jpeg</url>
            <title>Stories by Alexandr Wang on Medium</title>
            <link>https://medium.com/@alexandrwang?source=rss-97d94781bd42------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 02:20:45 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@alexandrwang/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Guest Post: Cloudinary Uses Scale API to Focus on Image Compression]]></title>
            <link>https://medium.com/scale-api/guest-post-cloudinary-uses-scale-api-to-focus-on-image-compression-a69a23bc3583?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/a69a23bc3583</guid>
            <category><![CDATA[image-processing]]></category>
            <category><![CDATA[api]]></category>
            <category><![CDATA[ycombinator]]></category>
            <category><![CDATA[scale]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Mon, 25 Sep 2017 21:02:10 GMT</pubDate>
            <atom:updated>2017-09-25T21:02:10.301Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*8Wxwy0l5qXKOiAtWWQiSmw.png" /></figure><p>We recently worked with the team at Cloudinary to help build and evaluate a better image compression and quality measurements. The results of Cloudinary’s work on Scale were very insightful and we wanted to share it broadly to demonstrate how more companies can leverage human judgments to build high quality features.</p><p>Without further ado, <a href="https://cloudinary.com/blog/author/tal_lev_ami">Tal Lev Ami</a>, CTO of Cloudinary, wanted to write about how they used Scale API:</p><p>Here at Cloudinary, we provide a cloud-based tool that enables our users to compress images and video for their websites and apps. Our goal is to preserve the visual integrity of the content, but deliver the smallest file size to any device or browser to ultimately optimize website performance and end user satisfaction.</p><p>One of the hallmarks of the Cloudinary solution is the ability to automate many functions of image compression, so that developers don’t have to spend time tweaking each photo and making multiple copies of different sizes and resolutions to fit every possible scenario. Compression algorithms can be tricky because they’re trying to make changes that have the smallest visual impact, but different images can react differently to compression.</p><p>As we were developing the algorithm for our “q_auto” capabilities — which strikes a balance between visual quality and file size — we needed to test it to understand how the resulting images compared to the human eye. Enter Scale API.</p><p>Many image compression formats — like JPEG 2000 and JPEG XR — have been tweaked to score well on particular metrics, such as peak signal-to-noise ratio (PSNR). But these don’t always correlate with human perception on image quality.</p><p>We leveraged Scale API to compare pairs of images and give us perspective on which image was liked most by humans. With Scale API, we did a variety of tests, comparing several formats, including WebP, JPEG 2000, JPEG XR (lossy) Lepton (MozJPEG, recompressed with Lepton), FLIF, BPG, Daala, and PNG8 (pngquant+optipng). We also were able to get input on the difference between the uncompressed original image vs. a compressed version.</p><p>Scale API enabled us to create A/B comparisons that were viewed by human observers. We submitted over 4,000 image comparisons to Scale API, sending at least four independent Scale API requests for each pair of image. This resulted in at least eight actual human comparisons for each pair of images. The outcome of these comparisons were evaluated beside other perceptual metrics such as PSNR, Google’s Butteraugli, DSSIM (Structural (Dis)Similarity) and a new metric Cloudinary developed called <a href="http://cloudinary.com/blog/detecting_the_psychovisual_impact_of_compression_related_artifacts_using_ssimulacra">SSIMULACRA (Structural SIMilarity Unveiling Local And Compression Related Artifacts)</a>.</p><p>The results showed that overall, PSNR is “correct” in only 67 percent of the cases. Butteraugli gets it right in 80 percent of the cases, and DSSIM in 82 percent of the cases. Our new metric, SSIMULACRA, agrees with human judgments in 87 percent of the cases. Looking just at the high-confidence human judgments, we found about 78 percent agreement for PSNR, 91 percent for both Butteraugli and DSSIM, and almost 98 percent agreement for SSIMULACRA. You can <a href="http://cloudinary.com/blog/detecting_the_psychovisual_impact_of_compression_related_artifacts_using_ssimulacra">read more about SSIMULACRA and these results on the Cloudinary blog</a>. Or if you want to give it a try: <a href="https://github.com/cloudinary/ssimulacra">SSIMULACRA is free and open-source software</a>!</p><p>The results of Scale API comparisons gave us useful data points to validate our metrics and provided more insights into the compression benchmarks we are running and the comparison of various image formats. And from these insights we were able to improve our visual perception metrics and fine-tune our “q_auto” functionality so we know how aggressively we can compress images.</p><p>Through this process we were impressed not only by the useful data points derived from the Scale API, but also the great support we got from the company and the product’s ease-of-use, all which came at a reasonable price.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a69a23bc3583" width="1" height="1" alt=""><hr><p><a href="https://medium.com/scale-api/guest-post-cloudinary-uses-scale-api-to-focus-on-image-compression-a69a23bc3583">Guest Post: Cloudinary Uses Scale API to Focus on Image Compression</a> was originally published in <a href="https://medium.com/scale-api">Scale API</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Introducing New Image Annotation Types]]></title>
            <link>https://medium.com/scale-api/introducing-new-image-annotation-types-47b0b482b7c2?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/47b0b482b7c2</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[api]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[self-driving-cars]]></category>
            <category><![CDATA[scale]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Wed, 16 Aug 2017 15:01:01 GMT</pubDate>
            <atom:updated>2017-08-16T15:01:01.457Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*z7kBmQxk12waaFBUCyfDNQ.png" /></figure><p>We’re excited to be launching a bunch of new annotation types for images. Since the launch of our bounding box API, we’ve annotated millions of images with boxes to identify a host of different objects, from cars and hats to roof damage and parking lots. Scale is becoming an industry-standard tool for solving computer vision problems.</p><p>We’re committed to offering developers all the tools they need. We’ve gotten many requests for a variety of image labeling types beyond boxes, and wanted to begin rolling out new labeling types to our entire community.</p><p>So, we’ll be releasing the following labeling types to all customers today:</p><ul><li>Pixel-wise Semantic Segmentation</li><li>Polygon Annotations</li><li>Point Annotations</li><li>Line Annotations</li></ul><h3>Pixel-wise Semantic Segmentation</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xAye9W5WESePwxbBybtEBw.png" /></figure><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/eabfe1c4b59741eb0db3d776b5d61f92/href">https://medium.com/media/eabfe1c4b59741eb0db3d776b5d61f92/href</a></iframe><p><a href="https://docs.scaleapi.com/#semantic-segmentation">Documentation</a></p><h3>Polygon Annotation</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yNDp84zNRuD0E-MrTmF5QA.jpeg" /></figure><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/8de9a05c402f17a05b6bef47643bfc76/href">https://medium.com/media/8de9a05c402f17a05b6bef47643bfc76/href</a></iframe><p><a href="https://docs.scaleapi.com/#polygon-annotation">Documentation</a></p><h3>Line Annotation</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_z-rYY2gKq7qN0G5nm_miA.png" /></figure><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/2c4069dd5a6c4e6bf07e68b9f500e7cf/href">https://medium.com/media/2c4069dd5a6c4e6bf07e68b9f500e7cf/href</a></iframe><p><a href="https://docs.scaleapi.com/#line-annotation">Documentation</a></p><h3>Point Annotation</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/1*vvczljGIqbcVxRj_zvEFGw.jpeg" /></figure><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/d47a16c2d252fb70b89aa2e47bf9132d/href">https://medium.com/media/d47a16c2d252fb70b89aa2e47bf9132d/href</a></iframe><p><a href="https://docs.scaleapi.com/#point-annotation">Documentation</a></p><p>If you have other feature requests or labeling types we should support, please reach out to us on our <a href="http://chat.scaleapi.com/">support Slack</a>! We currently have LIDAR point cloud annotation and cuboid annotation in beta.</p><p>You can get started with a free API key today!</p><p><a href="https://www.scaleapi.com/signup">Scale: API For Human Intelligence</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=47b0b482b7c2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/scale-api/introducing-new-image-annotation-types-47b0b482b7c2">Introducing New Image Annotation Types</a> was originally published in <a href="https://medium.com/scale-api">Scale API</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Word-by-word Timestamps for Audio Transcription]]></title>
            <link>https://medium.com/scale-api/word-by-word-timestamps-for-audio-transcription-7d50325689c0?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/7d50325689c0</guid>
            <category><![CDATA[speech-recognition]]></category>
            <category><![CDATA[transcription]]></category>
            <category><![CDATA[scale]]></category>
            <category><![CDATA[ycombinator]]></category>
            <category><![CDATA[api]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Wed, 31 May 2017 16:31:06 GMT</pubDate>
            <atom:updated>2017-05-31T16:36:04.525Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eSqEVCCl1rqXCKS205Fr2w.png" /></figure><p>Starting today, we will now be sending word-by-word timestamps for all of our audio transcription requests. It was a common feature request we’ve gotten for our audio transcription API, and has enabled a wide range of applications for our customers.</p><p>Successful transcriptions will now also include an alignment field, which will contain an array of time-aligned words (in the same order as the transcript), where each entry in the array has the following values:</p><ul><li>word: The word in question</li><li>start : timestamp in the audio file at which this word begins</li><li>end : timestamp in the audio file at which this word ends</li><li>confidence : The confidence for this word’s timestamps</li></ul><p>We are excited to see what developers decide to build using the alignment. We’ve seen customers already build custom captioning, simultaneous text-and-video editing, and more.</p><p>Check out our <a href="https://docs.scaleapi.com/#callback-format18">docs</a> to learn more! And you can sign up for an account below, and we’ll get you started with 5 free API requests!</p><p><a href="https://www.scaleapi.com/signup">Scale: API For Human Intelligence</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7d50325689c0" width="1" height="1" alt=""><hr><p><a href="https://medium.com/scale-api/word-by-word-timestamps-for-audio-transcription-7d50325689c0">Word-by-word Timestamps for Audio Transcription</a> was originally published in <a href="https://medium.com/scale-api">Scale API</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Announcing our Series A with Accel]]></title>
            <link>https://medium.com/scale-api/announcing-our-series-a-with-accel-b467757c9a47?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/b467757c9a47</guid>
            <category><![CDATA[api]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[developer-tools]]></category>
            <category><![CDATA[scale]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Wed, 24 May 2017 01:27:39 GMT</pubDate>
            <atom:updated>2017-05-24T04:42:39.911Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*94NiroFaZEpVxmN61p_slQ.png" /></figure><p>We are excited to <a href="https://www.wsj.com/articles/scale-helps-engineers-outsource-microtasks-1495452603">announce</a> that we have raised a $4.5 million Series A round of funding led by Accel. Along with this funding, Accel’s Daniel Levine has joined Scale’s board. The funding will be used to invest in our rapid growth, expand our offerings, and grow our team.</p><p>When we started Scale ten months ago, our goal was to build the simplest API for human intelligence. Since then we have released six APIs (<a href="https://www.scaleapi.com/image-annotation">Image Annotation</a>, <a href="https://www.scaleapi.com/audio-transcription">Audio Transcription</a>, <a href="https://www.scaleapi.com/ocr-transcription">OCR Transcription</a>, <a href="https://www.scaleapi.com/categorization">Categorization</a>, <a href="https://www.scaleapi.com/comparison">Comparison</a>, and <a href="https://www.scaleapi.com/data-collection">Data Collection</a>), and we process millions of API requests each month.</p><p>This success has confirmed our conviction that there is no better time than now to build Scale. We are currently at an inflection point in the age of artificial intelligence (AI). Just like software before it, AI will have an enormous effect on productivity and enable incredible new technologies. Scale works with many companies and technologies at the forefront of this shift, including self-driving cars, intelligent assistants, and voice analytics. Our customers agree that integrating AI with accurate human intelligence is crucial to building reliable AI technology. As a result, we believe Scale will be a foundation for the next wave of development in AI.</p><p>This financing will allow us to grow our engineering, operations, support, marketing, and sales teams to continue on our journey to make our customers more productive and efficient. We’re committed to continuously improving our costs, speed, and quality and to providing a quality customer experience.</p><p>We’d also like to thank our colleagues, customers, partners, and investors for supporting on this journey to build APIs for human intelligence. We look forward to continuing to support companies and developers in creating innovative software in the age of AI. This is just the beginning.</p><p>We are actively hiring exceptional engineers, designers, AI researchers, and more. If you are interested in joining Scale, check out our <a href="https://www.scaleapi.com/about#jobs">careers page</a>.</p><ul><li><a href="https://www.scaleapi.com/about#jobs">Scale: API For Human Intelligence</a></li><li><a href="https://www.wsj.com/articles/scale-helps-engineers-outsource-microtasks-1495452603">Scale Helps Engineers Outsource &#39;Microtasks&#39;</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b467757c9a47" width="1" height="1" alt=""><hr><p><a href="https://medium.com/scale-api/announcing-our-series-a-with-accel-b467757c9a47">Announcing our Series A with Accel</a> was originally published in <a href="https://medium.com/scale-api">Scale API</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What I learned in 2016]]></title>
            <link>https://medium.com/@alexandrwang/what-i-learned-in-2016-5526cff37035?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/5526cff37035</guid>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[mit]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[reflections]]></category>
            <category><![CDATA[ycombinator]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Sat, 31 Dec 2016 21:13:05 GMT</pubDate>
            <atom:updated>2018-10-23T22:38:34.984Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-ND4cIweE4KOJ3AOk_iFKQ.png" /></figure><p>2016 has been an incredible year for <a href="http://www.scaleapi.com/">Scale</a>, a company I started in June. We’re helping amazing companies like Uber, Alphabet, Houzz, and P&amp;G with our API for human intelligence. Our mission to bring human intelligence to software applications aligns with what Eric Schmidt believes will <a href="http://www.newsweek.com/eric-schmidt-predicts-next-100-billion-company-471120">shape the next $100B company</a>, and more importantly activate one of the greatest technological shift of our time. I sent my last investor update with a sense of disbelief that Scale was the same company we started back in June.</p><p>When reflecting on my year, I felt defined by two things: adopting a risk-seeking mentality and doing a lot of hard work. I’ve gained a little perspective on the importance of both, and wanted to reflect on my experiences before the new year.</p><p>While we were fundraising for Scale, one VC asked me:</p><blockquote>When did you know you wanted to start a startup? The past few years of your life seem perfectly lined up to start your own startup.</blockquote><p>I remember being caught off guard, and rattled off some silly answer about being inspired by <em>The Social Network.</em></p><p>The earliest I can remember wanting to start a company was when I was in 9th grade. My friend and I started a Google Doc of startup ideas to do when we were old enough. It contains gems like “superpopular app. any ideas?”, “iTunes replacement (I swear it always freezes)”, and “actually nvm spotify is pretty dope”.</p><p>However, I haven’t lived the past few years with the sole goal of starting a startup. Instead, I’ve relentlessly pursued opportunities any I found exciting, consuming, and time-sensitive. Put more simply, I’ve been <strong>minimizing my regret, </strong>an <a href="https://en.wikipedia.org/wiki/Regret_(decision_theory)">important concept</a> in decision theory and machine learning.</p><p>In December 2015 while a freshman at MIT, I was interviewing everywhere, perhaps indiscriminately, for engineering and machine learning opportunities which excited me, from Deepmind and Snapchat Research to Opendoor and Stripe. On one hand, these opportunities really excited me. But at the same time, I felt aimless and a little lost—I was waiting for one of these offers to feel irrefusable, but it really felt like I was going to go work at yet another tech company. Having already worked at a few mid-stage startups after high school, it didn’t feel like a growth opportunity for myself.</p><p>While discussing my offer with Eric Wu, <a href="http://opendoor.com/">Opendoor</a>’s CEO, he offered an insight that really struck me. It went something like this:</p><blockquote>Think about it this way. As you get older, you’re only going to become more and more risk-averse. You’ll get married, buy a house, have kids, etc.</blockquote><blockquote>That means today, you’re the most risk-seeking you’ll ever be. <strong>So, if there’s a risk today that you aren’t willing to take, then you might never end up taking that risk.</strong></blockquote><p>At the time, he was trying to persuade me against larger companies like Stripe or Dropbox in favor of a smaller company like his own Opendoor (which at the time was around 30 people). But overall, it persuaded me against accepting any offers at all.</p><p>Instead, the conversation convinced me that I needed to try to start a company. Not only would it be a great growth opportunity, I knew I would regret it if I never took the risk to be an entrepreneur at the perfect time.</p><p>I knew I would also regret not accepting some of the offers I received; Snapchat and Opendoor are doing particularly well today. There are obviously so many opportunities in technology, and the outcomes have such high variance, that it‘s nearly impossible to not regret some of your choices. However, I believe that if you don’t have opportunities you regret not taking, you aren’t chasing after enough opportunities.</p><p>That was the start of my journey that leads to me founding Scale. In January, while I was working as at Hudson River Trading, a high-frequency trading firm, I began hacking on side projects. I went through a couple ideas over the next few months, consistently learning and iterating. When I finally landed on Scale, it felt like magic. The response from our <a href="https://www.producthunt.com/posts/scale">Product Hunt launch</a> and <a href="https://techcrunch.com/2016/07/25/scale-lets-companies-outsource-their-human-powered-tasks-with-one-line-of-code/">TechCrunch article</a> was incredibly positive, and we’ve been growing rapidly ever since.</p><p>Over the summer, we were a part of the Y Combinator (YC), a startup accelerator known for producing great companies like Airbnb, Dropbox, and Stripe. Something the YC partners tell you at the beginning of the program is that you should always be working extremely hard. Your time during YC is very highly leveraged—progress made during the summer directly translates into easier fundraising at Demo Day, which might be the single best fundraising environment out there. Paul Buchheit, one of the YC partners, even said that many consider YC the most productive time of their lives.</p><p>I had already concluded before YC that the amount I learned grew super-linearly with how hard I worked. That is, there’s an economies of scale effect with how hard I worked. After my first few months of working 12-hour days at Quora, I remember being really surprised at how much I’d improved as an engineer. It felt like I went from a code monkey to a legitimate system architect in just a few months, even though I had been coding for years beforehand.</p><p>Looking back on the year, it still rings true. For the first third of the year while I was still a student at MIT, I was juggling 5 graduate computer science courses, along with the side projects I was working on, which consumed most of my attention. I didn’t take any breaks—I spent my spring break building the iOS app for <a href="http://www.heyava.com/">Ava</a>, a precursor to Scale, and flew out to SF right after finals to work on it. I remember thinking many times throughout that semester that I was working harder than at any other point in my life.</p><p>I felt the exact same way during YC. It turns out that if you have a lot of momentum, it’s not too difficult to keep grinding even further. In <a href="http://paulgraham.com/corpdev.html">one of Paul Graham’s essays</a>, he says one of the tricks to surviving startups is to not stop to think about how tired you are. After a full year of 80–100 hour weeks, I feel like I must have employed that trick.</p><p>Generating a lot of ideas is another great motivator for hard work. Oftentimes, makers will find themselves in situations where the rate at which they get ideas greatly exceeds the rate at which they can build them. Even better is when the rate of <strong>good</strong> ideas exceeds the rate at which they can build them. At this point, time becomes the enemy, and you begin craving 100 hour weeks. The higher your execution speed, the more you’re able to keep up with the flood of ideas.</p><p>This flooding (and sometimes wastage) of ideas is one of the reasons why moving fast is so important for startups. The return on investment on most ideas in startups is distributed roughly according to a power-law distribution (similarly to startup outcomes themselves). There are a few ideas, like Snapchat Stories or Facebook News Feed, whose return will greatly overshadow the other ideas. Prioritization seeks to identify these ideas with outsized returns as best as possible, but it’s impossible to prioritize perfectly. To prevent a diamond idea from being passed over, startups need to compress each idea to an minimum viable product (MVP) as much as possible, and focus on fast execution speed.</p><p>That’s why we focus on moving fast at Scale—it’s something that the whole team and I have truly internalized. I’m immensely proud of our team for continuing to ship <a href="https://www.scaleapi.com/image-recognition">feature</a> after <a href="https://www.scaleapi.com/audio-transcription">feature</a> after <a href="https://blog.scaleapi.com/introducing-our-data-collection-endpoint-729ced2c681a">feature</a> after <a href="https://blog.scaleapi.com/scale-api-python-client-library-a90d0e0bb041">feature</a>, and it bodes incredibly well for everything we’ll do at Scale in the future. Hard work consistently seems to trump other factors.</p><p>Ultimately, 2016 might go down as the year I personally worked the hardest. Although, I surely hope not—it’s an incredible feeling to be in the midst of the most productive time of your life so far.</p><p>By the way, we’re <a href="https://www.scaleapi.com/jobs">hiring</a> for engineers and operations! Feel free to apply online, or simply email hello[at]scaleapi.com.</p><p><em>Thank you to </em><a href="https://medium.com/u/dc5076cb445a"><em>Leigh Marie Braswell</em></a><em> and Jenny Wang for reading earlier drafts.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5526cff37035" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Introducing our Audio Transcription API]]></title>
            <link>https://medium.com/scale-api/introducing-our-audio-transcription-api-a9180cbf5ea7?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/a9180cbf5ea7</guid>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[api]]></category>
            <category><![CDATA[transcription]]></category>
            <category><![CDATA[audio]]></category>
            <category><![CDATA[scale]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Tue, 06 Dec 2016 23:09:40 GMT</pubDate>
            <atom:updated>2016-12-06T23:11:25.865Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mSEp45fYJzgX5z049GQJuw.jpeg" /></figure><p>We are really excited to release our <a href="https://www.scaleapi.com/audio-transcription"><strong>audio transcription API</strong></a> today to developers everywhere!</p><p>We have heard many times from developers that they need better tools to make sense of their audio data. Just like when we launched our <a href="https://blog.scaleapi.com/introducing-our-bounding-box-endpoint-7f2b1d07f7a7#.4anc516dj">image annotation API</a>, we recognize that audio comprehension is one of those things humans are fundamentally better than machines at. A powerful audio transcription API is an important step in empowering developers with that human intelligence.</p><p>Our API for audio transcription allows developers and companies utilize Scale’s combination of AI and human-in-the-loop processes to transcribe entire corpuses of audio, cost-effectively at scale.</p><p>We do all the heavy-lifting in the background, so you can get started with just a few lines of code:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/a2146995ab6ba8169a16037aeef12b46/href">https://medium.com/media/a2146995ab6ba8169a16037aeef12b46/href</a></iframe><p>We’re really excited to see what developers decide to build on top of our API. Here are just a few ideas to get you started:</p><ul><li><strong>Transcribing Phone Calls:</strong> Your businesses could be generating millions of minutes of phone call recordings every day, but it can be incredibly difficult to get any useful information just from the audio files. Transcribe those phone calls to quickly scan and run analytics on the resulting transcripts.</li><li><strong>Audio Search Indexing:</strong> Imagine a search engine for the world’s audio files. Without any processing on the audio files, such a task would be incredibly difficult. With our audio transcription API, it’s possible to create such a search engine with existing techniques on the text of the transcripts.</li><li><strong>Automated Meeting Transcripts: </strong>Automated meeting notes become incredibly simple using our API, and could greatly accelerate the productivity of your company.</li><li><strong>Transcribing Voicemail and Podcasts:</strong> Listening to audio can be incredibly time-consuming. Transcribe the audio for the ability to quickly consume information from voicemail, podcasts, or just about anything else.</li><li><strong>And more..</strong></li></ul><p>We hope you are as excited as us about this launch. This is just the beginning and our team will continue to add more functionality as we listen to feedback from our developers. Some of the features we intend on adding in the coming weeks:</p><ul><li><strong>Speaker recognition for multi-party conversations. </strong>Transcribing phone calls and meetings is an important use case for us to handle, and we want to enable those use cases.</li><li><strong>Timestamps with transcripts. </strong>In the same vein, captioning is an important use case for audio transcription. We’re really excited to enable developers to build custom captioning technology using our API.</li></ul><p>If you have any questions, we are here for you — email <a href="/service/mailto:hello@scaleapi.com">hello@scaleapi.com</a> or join our <a href="https://chat.scaleapi.com/">Slack</a>.<br> <br>We are looking forward to see what you create on top of our API.</p><p><a href="https://dashboard.scaleapi.com/signup">Sign up</a> for an API key today!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a9180cbf5ea7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/scale-api/introducing-our-audio-transcription-api-a9180cbf5ea7">Introducing our Audio Transcription API</a> was originally published in <a href="https://medium.com/scale-api">Scale API</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Scale API Python Client Library]]></title>
            <link>https://medium.com/scale-api/scale-api-python-client-library-a90d0e0bb041?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/a90d0e0bb041</guid>
            <category><![CDATA[python]]></category>
            <category><![CDATA[ycombinator]]></category>
            <category><![CDATA[scaleapi]]></category>
            <category><![CDATA[scale]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Wed, 19 Oct 2016 03:48:33 GMT</pubDate>
            <atom:updated>2016-10-19T03:48:33.006Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IcBOpk1z-iAe7n25ymzm-w.png" /></figure><p>We’re excited to launch a <a href="https://github.com/scaleapi/scaleapi-python-client/">Python client library</a> for Scale API!</p><p><a href="https://github.com/scaleapi/scaleapi-python-client/">scaleapi/scaleapi-python-client</a></p><p>Installation is crazy simple, and you can get started right now by viewing our <a href="https://docs.scaleapi.com/?python">Python docs</a>.</p><p>$ pip install scaleapi</p><p>As an API company, our focus is on creating a stellar developer experience. Python is one of the <a href="http://pypl.github.io/PYPL.html">most popular programming languages</a> right now, and we want to make it as easy as possible to use Scale.</p><p>As always, feel free to <a href="https://chat.scaleapi.com/">chat us on Slack</a> for comments and feedback!</p><p>Sign up for a free API key today to get started! We’re eager to see the amazing things you build :)</p><p><a href="https://dashboard.scaleapi.com/signup">Signup for Scale</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a90d0e0bb041" width="1" height="1" alt=""><hr><p><a href="https://medium.com/scale-api/scale-api-python-client-library-a90d0e0bb041">Scale API Python Client Library</a> was originally published in <a href="https://medium.com/scale-api">Scale API</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Markdown in Task Instructions]]></title>
            <link>https://medium.com/scale-api/markdown-in-task-instructions-7d7678a20867?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/7d7678a20867</guid>
            <category><![CDATA[scale]]></category>
            <category><![CDATA[markdown]]></category>
            <category><![CDATA[scaleapi]]></category>
            <category><![CDATA[ycombinator]]></category>
            <category><![CDATA[slack]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Mon, 17 Oct 2016 20:42:17 GMT</pubDate>
            <atom:updated>2016-10-17T20:47:52.240Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*N6DnnBsxWSQfiocR_anlFg.png" /></figure><p>Starting today, you can now use <a href="https://github.com/adam-p/markdown-here/wiki/Markdown-Cheatsheet">markdown</a> in your Scale API requests!</p><p>Through the past few months, we have learned a lot about tasks. Some tasks are simple, like determining whether an image has a dog. Others are more complicated, like moderating content according to the website’s unique guidelines.</p><p>One limitation we’ve found is that plaintext instructions were not enough. It can be incredibly helpful to add <strong>emphasis</strong>,</p><ol><li>lists</li><li>of</li><li>instructions</li></ol><p>and images to instructions.</p><p>More detailed, feature-rich instructions results in better communication, which results in more clarity for our Scalers and higher quality for our users. It is really a win-win.</p><p>Check out our <a href="https://docs.scaleapi.com/">docs</a> for all the information. If you have any questions or feedback (about this change or Scale in general), please chat with us on Slack!</p><p><a href="https://chat.scaleapi.com/">Join Scale API Support on Slack!</a></p><p>And sign up for a free API key today! You won’t regret it.</p><p><a href="https://dashboard.scaleapi.com/signup">Signup for Scale</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7d7678a20867" width="1" height="1" alt=""><hr><p><a href="https://medium.com/scale-api/markdown-in-task-instructions-7d7678a20867">Markdown in Task Instructions</a> was originally published in <a href="https://medium.com/scale-api">Scale API</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What I wish I had known before YC]]></title>
            <link>https://medium.com/@alexandrwang/what-i-wish-i-had-known-before-yc-8e0edcf318b9?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/8e0edcf318b9</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[ycombinator]]></category>
            <category><![CDATA[mit]]></category>
            <category><![CDATA[founders]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Sun, 25 Sep 2016 18:21:13 GMT</pubDate>
            <atom:updated>2016-09-26T02:29:30.457Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*brI8nD9iWky5q2cGXP_RAA.jpeg" /></figure><p>Before I went through YC, I thought that it would transform me into an incredible entrepreneur. No matter who I was when I went in, if I worked hard, I would emerge an all-knowing YC graduate. This is how we’re traditionally supposed to think of schools (e.g. MIT will turn me into an amazing engineer), and for better or for worse, I thought of YC as the premier “startup school”.</p><p>This, of course, isn’t true. The truth is, you will come out of YC as more or less the same person (this is what the title refers to—but allow me to explain). YC will teach you many things, but nothing <em>transformative</em> (which, by the way, most people would call “culty”). It will “teach” you to work harder, talk to your users more, make something people want, and then 10 other things specific to your business.</p><p>But it won’t “teach” you how to be the next Elon Musk—nobody can teach you that. If somebody could teach you, that would imply it could be reliably replicated, which would make Elon Musk and his ideas worth significantly less than they are in practice. If you are “the next Elon Musk”, it’s because of who you were before YC, and you probably won’t even realize it until you’ve worked really fucking hard for years and years.</p><p>Here’s YC’s value to the market: identifying really incredible, brilliant, driven founders who are tackling valuable problems. <em>That</em> is why YC itself makes money (and all VCs by the way). They have hypotheses about valuable ideas (which they <a href="http://www.ycombinator.com/rfs/">publish</a>, and can be convinced on too), and are <strong>really</strong> <strong>really</strong> good at finding great people. I can personally attest—my S16 batchmates are an extraordinarily talented bunch who work their asses off.</p><p>After that, YC puts you in the best position imaginable for a young startup founder. They tell you what to focus on (namely your users). They will tell you when what you’re doing isn’t working (sometimes by just pointing to your traction). They will point you to helpful alumni to talk to about the smallest facets of your business. <strong>But the rest—that’s all you and your hard work.</strong></p><p>For me, that’s reassuring. It means that what stands between me and a billion dollar company isn’t more YC pixie dust, but more hard work just as I had been doing before YC. Looking back, I can see why our clubbing app Listed failed, or why Ava didn’t work out, but I only learned that by doing it. <strong>That’s the secret</strong>—you’ll only learn things by doing them, especially in startups. YC can save you some steps because they’ve seen it before, but you still need to move fast and do things.</p><p>If you’re working on a valuable idea, <a href="http://www.ycombinator.com/apply/">apply to YC</a>. You will regret not applying (I did once or twice). And if you have any humans-in-the-loop with your startup, check out <a href="https://scaleapi.com/">Scale</a>, and ping me on <a href="https://scalesupport.herokuapp.com/">Slack</a> with any questions.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8e0edcf318b9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Scale Annotation Demo]]></title>
            <link>https://medium.com/scale-api/scale-annotation-demo-a996c7c84373?source=rss-97d94781bd42------2</link>
            <guid isPermaLink="false">https://medium.com/p/a996c7c84373</guid>
            <category><![CDATA[api]]></category>
            <category><![CDATA[image-recognition]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[scale]]></category>
            <category><![CDATA[ycombinator]]></category>
            <dc:creator><![CDATA[Alexandr Wang]]></dc:creator>
            <pubDate>Tue, 20 Sep 2016 18:20:12 GMT</pubDate>
            <atom:updated>2016-09-20T18:20:12.058Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pzkY8i_gft8xFsRz1HiL0w.png" /></figure><p>We want to make it easy to use Scale, and we’re really excited about our new <a href="https://blog.scaleapi.com/introducing-our-bounding-box-endpoint-7f2b1d07f7a7">annotation API</a>. We just can’t wait to see the great things you build using it.</p><p>So, we wrote a little demo (written in Ruby) which demonstrates the power of our annotation API, and open sourced it so you can build off of it.</p><p>You can start playing with it on <a href="http://annotate-demo.scaleapi.com/">our hosted demo</a> right now:</p><p><a href="https://annotate-demo.scaleapi.com/">scale-annotate-demo</a></p><p>If you’re the hacker type, you can <a href="https://github.com/scaleapi/scale-annotate-demo#deploy-on-heroku">deploy your own instance to Heroku</a> in minutes. Below is our repository for more information:</p><p><a href="https://github.com/scaleapi/scale-annotate-demo">scaleapi/scale-annotate-demo</a></p><p>We hope you have hacking on it as we did building it. <a href="https://dashboard.scaleapi.com/signup/">Sign up</a> for Scale today to start hacking with some API keys!</p><p>As parting words, here’s the Ruby code we used in our demo to integrate with Scale:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/6f3c91965d0c0d01fee6b8251afe0b86/href">https://medium.com/media/6f3c91965d0c0d01fee6b8251afe0b86/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a996c7c84373" width="1" height="1" alt=""><hr><p><a href="https://medium.com/scale-api/scale-annotation-demo-a996c7c84373">Scale Annotation Demo</a> was originally published in <a href="https://medium.com/scale-api">Scale API</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>