I spent over 6 years at Google, and these were the best years of my professional life. I learned volumes from my brilliant colleagues, got appropriately humbled and was lucky enough to interact and sometimes closely work with such champions of the industry as Jeff Dean, Peter Norvig, Amit Patel, Marissa Mayer and the like. I also got to experience the incredible rush of releasing the newborn Google Trends to judgment of millions of users and the press.
But all good things come to an end, and in April 2011 I left Google to join a little known data startup called Factual. Headquartered in Los Angeles, they wanted to expand their presence in the Bay Area, I loved their mission and I knew the founder, Gil Elbaz, personally from the times where he was heading Google’s Santa Monica office after it acquired Applied Semantics, which he also had founded. It was a great opportunity to learn the ropes before starting my own company, and Gil and I agreed I might not be staying for too long.
The very first thing I learned at Factual is that the startup wilderness is nothing like Google. The Google’s technology stack is amazing. Not only it’s highly advanced, but also well documented, well supported and constantly improving. In contrast, out there we rely on open source projects and 3rd party services, and to my surprise it turned out that quite a number of things I used to take for granted are either impossible or quite complicated to do. And often there’s no support to speak of.
But this only makes things more interesting, and I had a great time at Factual – it definitely has its share of fascinating people. At Factual, I got to lead the creation of Factual’s first product database. And there was a lot to learn, too: I had a quite unique experience of immersing myself in Clojure – a JVM-based Lisp. I had never coded practically and extensively in Lisp, and now I know that it really is true that Lisp rewires your brain. I can confidently say that it made me a better engineer.
Without doubt, the most rewarding project at Factual was the workflow management tool Drake, which we just recently released to open source. Drake allows to organize and automate off-line data processing pipelines by specifying steps and their dependencies, in a similar way make automates building software. There are quite a few features which makes Drake stand out, such as multiple outputs, HDFS support or precise control over step execution. I really wish I had a tool like Drake when I was working on Google’s QueryCount service back at 2004. Instead I had to write a lot of Python code to automate my data workflow and I hated it. Based on the response we’ve been getting so far, it seems like we hit a nerve – and it’s exciting to watch our small Drake community grow. Needless to say, Drake is written in Clojure.
But as it happens, I never got over consumer products. My true passion has always been making something millions of people would use. So, about a month ago, I quit Factual to found a new company in the consumer space, based on the idea I’ve been thinking about for a couple of years now. I’d love to tell you what it is but then I’d have to kill you. And I’m pretty sure I need you alive because I myself am dead set on building a small team of world-class engineers and doing all I can to make it happen.
If you’re an engineer or a UI designer looking for an exiting project that has the potential to change the way we do certain things, let me know! My email is firstname.lastname@example.org, and I would love to hear from you. And of course, if you’re an angel investor or a VC searching for opportunities, I’d be more than happy to get together.
I was told it won’t be easy but I’m braced for the ride. Wish me luck.