On life before computers

I grew up in Bangalore, India, back when it was a sleepy little town. Folks picture Bangalore as a tech cluster today, but it wasn’t like that yet. For the longest time I wanted to be a theoretical physicist and was dimly aware of computers.

In middle school we couldn’t afford a computer at home, so I learned to program on pen and paper at a bootcamp in the neighborhood one summer. Once you got your work reviewed by the instructor, you got the chance to type it into the few machines they had in the basement. I only went to the bootcamp because I was a nuisance at home and my parents needed to get me out of the house.

As I got closer to college I wanted out of the “straight and narrow” path of an engineering education, and at the time that felt like the only option. I really felt like the theoretical physics research career I wanted was incompatible, so I wound up finding my way to Williams College in the United States, pursuing a liberal arts education with a blank slate. For a while I was leaning towards a history and economics education, but figured it was prudent to have one grown up major in the mix, and ironically, that's when I discovered computer science was an actual field, and fell in love completely.

On getting too much education

One downside of a broad liberal arts education is that you feel like you just don’t have any depth whatsoever, and I felt like I wanted a lot more computer science, so I plunged into a PhD in computer science straight away.

A PhD is interesting, because the north star is to become a tenure track faculty member teaching whatever subfield you’re in. But I knew quickly that I didn’t want to be an academic. I was constantly on the hunt for projects that I felt would have real-world impact, as opposed to getting me published. I ended up playing around with theorem provers, type systems, databases, deterministic runtimes, compilers, and distributed systems. Overall, a fantastic computer science education.

On spending a decade in databases

About halfway through my PhD, once I felt I had a solid foundation underfoot, I went hunting for the most impactful projects, and was very taken in by a few. The first was the Spanner paper, in 2012. I felt strongly at the time that the way forward was for distributed systems projects to stop exposing all their warts to users, and become more fully featured databases. The zeitgeist at the time was that databases were history, and the future was unbundling things. Being in the midst of it all, I found myself very much in disagreement, and found my way to Cockroach Labs shortly thereafter.

The second project I was obsessed with was the Timely Dataflow project at Microsoft Research. I spent a lot of time convincing Frank McSherry to commercialize the technology, one thing led to another, and we cofounded Materialize together. We built a database with a stream processor at its heart, delivering incrementally updated materialized views. 

Databases are hard. The saying goes that it takes 5-7 years to build a decent one, and that’s if you’re lucky. But it’s real in a way that a lot of projects today aren’t; you’re confronting some of the most thorny problems in technology completely head on.

On working with founders

A few years into Materialize, I found myself an angel investor in many fantastic companies, a good several of them where Amplify had invested as well. I found that I enjoyed helping other founders a lot. More than I thought I would at first. What really gives me energy is talking to technical founders building technical companies.

Starting a company is very hard and very lonely. And it’s tricky when you’re on the outside, because the reality is that you can do a lot more harm than good. When I call my founder friends, Amplify is consistently the place that we all felt knew technical founders the best. It feels like home.

On the Roman Empire

I feel very misunderstood, I’m more of a late republic guy, I really don’t think about the empire that much.