Why I’m leaving Google for a startup

A long-time Googler jumps out of the frying pan and into the fire.

After more than eight and a half years as an engineer and manager at Google, I’ve decided to tackle a new challenge by joining XNOR.ai, an early-stage startup developing AI for embedded devices. I’m equal parts excited and scared — it feels a bit like I’m playing the Bird Box Challenge with my career.

Although taking a new job usually does not involve protecting your children from unseen sky demons while rowing a boat blindfolded. I hope.

Ironically, it was also after about eight years at Harvard that I left my faculty position there to join Google. So the timing seems right to make another big career change.

A quick recap on the story so far

I joined Google back in 2010, initially on sabbatical from Harvard, and after a few months decided to stay on full time. When I moved to Seattle in 2011, my charter was to build a mobile web performance team within Google, a counterpart to the (then) desktop-focused Make The Web Fast effort. I was given a chunk of headcount and pretty much free rein to figure out what to do. We hired a bunch of great people, did a lot of benchmarking and data analysis, and prototyped various ideas (some good, some not so good). Eventually (and I am compressing time a lot here) we built the Flywheel proxy service to optimize mobile web pages, which was launched as Chrome Data Saver. Flywheel now has more than 700 million users (!) and is far more successful than I could have ever imagined.

Over the next few years, my team expanded to cover the majority of Chrome’s “Next Billion Users” efforts — developing Chrome features specifically targeting users in countries such as India, Indonesia, and Nigeria. The team now comprises about 40 engineers across four subteams, not counting all of the awesome PMs, UX designers, researchers, and test engineers. We added support to Chrome for saving, reading, and sharing offline pages for people with intermittent Internet access; added the “articles for you” suggestions to help users find more web content; developed a range of both server- and client-side optimizations to make web pages load faster on slow networks. We grew Chrome on Android to more than a billion users in our target countries. I am incredibly proud of all that we accomplished, and will look back on these years as among the happiest and most productive of my career.

So, things have been going really well. Why the heck would I leave?

I want to make one thing clear — I’m pretty happy at Google. Despite the many recent controversies surrounding the company, those things aren’t driving me away. (Although they do drive me nuts.) Google’s a place with tremendous opportunities to learn, work with smart people, and make a real dent in the world. There are definitely things I would change about Google, but on the whole it’s been fantastic and I feel so lucky to have had the opportunity to work there.

Missing the boat

My decision to jump ship came down to one simple thing: I wanted to challenge myself by doing the startup thing.

I’ve had a few opportunities to work at startups in the past, but always chickened out. Back in 2002, I had an offer to join a startup called Google, where I would have been employee number 400 or something. At the time, they were “just another search engine”— this is before Gmail, before Android, before Chrome, and of course before their IPO. But I was focused on going into academia and did not recognize the opportunity that Google presented at the time. After all, Google was basically doing the same thing as Alta Vista, and I had a hard time seeing how Yet Another Search Engine was going to change the world.

I also missed the boat when it came to Facebook. Back at Harvard, I had tried to talk Mark Zuckerberg (who was in one of my classes) out of leaving school to start the company, thinking it was too similar to MySpace and Orkut. In 2010, I had an opportunity to join them, again pre-IPO, but at the time was happily ensconced at Google and had little reason to leave.

Updating my utility function

My personal utility function has always assigned a much higher value to the intellectual content of the work that I’m doing than to the potential for financial reward. As a result, my instincts have always been to play it safe, leading me to prefer a more stable career move over a riskier one.

This time around, I feel that I’ve reached a stage in my career where I can afford to take those risks. I’ve done the professor thing, and been awarded tenure. I’ve done the big company thing, and built a large, successful team. Taking a foray into the startup world isn’t going to impact my career trajectory in any negative way, even if the company goes under — indeed, I think I need to get out of my comfort zone if I’m going to grow at all.

This is an opportunity to have substantial impact on a small company early in its lifecycle. If Google is a 100,000-ton aircraft carrier (with free snacks and massage chairs), a startup is like a 30-foot sailboat — far more nimble, but also much more exposed to the elements. There’s also a kind of energy and scrappiness in a startup environment that really appeals to me, like getting back to my research lab days.

The other thing is that I’m tremendously excited about the technology that XNOR.ai is developing. I’ll blog more later about my decision to join this specific company, but the work they are doing is very much up my alley. I’m an embedded and mobile systems hacker at heart, and being able to use my existing skills while learning about a new area hits the sweet spot for me.

We’ll see where this leads. And to all of my friends and colleagues at Google, thank you for all of the great times — I’m excited to see what you do next.