What do you get when a few (dozen) quantum mechanics Ph.D.s get together to build a system that brings Google Analytics-level usability to the hyper-complex world of financial trading?
You get Kensho, the Boston-based data analytics and machine intelligence company acquired by S&P Global in April of this year. In just three years, the company has snagged top spots on fintech watchlists from Fortune and Forbes, earned a $500 million valuation, and gone from accelerator to fully realized product — one that’s integrated into some of the weightiest trade desks in the world.
We caught up with David Relyea, who heads up a growing machine learning and artificial intelligence team out of Kensho’s brand new Pacific Palisades office, to hear more about the company’s recent LA move, the team he’s building and what they’ve got in store next.
Kensho has offices in Cambridge, D.C. and New York City. What’s the company hoping to tap into by laying down roots in Los Angeles?
Los Angeles is an unusual region. It has the single largest graduating pool of STEM researchers in the U.S. every year, yet its tech scene is much smaller than the San Francisco Bay Area or New York City. That enormous pool of talent makes it the perfect place for a new or growing company.
The team at Kensho is extremely impressive — there’s no shortage of Ph.D.s on your roster. Outside of pedigree, what’s one thing all Kensho employees have in common?
Every employee at Kensho is driven (in a somewhat low-key way) to have outsized impact and perform at a super high level. It’s what naturally happens when you get a lot of Ph.D.s from top schools together. That, and a lot of board games.
At the same time, people here are humble, mostly because everyone here can find several experts in whatever they're trying to learn.
A lot of companies don’t give much attention to how collaboration happens with remote teams or geographically diverse offices. At Kensho, it’s something we prioritize.”
With offices in multiple time zones, what does collaboration look like day-to-day?
A lot of companies don’t give much attention to exactly how collaboration happens with remote teams or geographically diverse offices. At Kensho, it’s something we prioritize.
We make sure everyone stays in close contact at all levels (not just manager and direct-report) between all of our offices, and that people are talking regularly. This ensures that communication flows well between everyone and that we’re consistently prioritizing the most important projects and sharing knowledge across the company. It also reduces everyone’s blockers and improves efficiency.
How would you describe your style as a manager?
When I was an individual contributor, my entire mindset was to optimize everything as much as possible. (The number of times I’ve rewritten this paragraph should attest to that.) When I became a manager, my entire focus shifted. Now my mindset is to grow every member of this group as quickly as possible. That’s my number one priority above everything else.
My reasoning is simple: A lot of managers are advised to speak to their employees with the goal of maximizing their contributions to the company. That’s bad, because it can lead to really weird behavior from the manager, as well as create a strangely superficial relationship between the two.
My mindset is to grow every member of this group as quickly as possible. That’s my number one priority above everything else.”
Growing employees is a much better thing to do, because it makes everyone happier and more productive. It’s also easier, because most people want to grow and further their career. I’ve heard other managers suggest that growing employees will just lead to them leaving faster, and I’ve found the exact opposite to be true.
If someone wants to grow and learn and has a happy place to do so, they’ll usually stick around. And if someone ends up outgrowing their role — great! Either the company benefits by giving them more responsibility, or the employee benefits by moving on and finding a bigger and better role elsewhere.
What specific skills or attributes are you looking for in a future hire?
We’re generally interested in two kinds of candidates. The first is someone fresh out of a Ph.D. or postdoc program, usually an expert in a specific area, really smart, persistent and with a desire to learn as much as possible. This first kind of candidate typically needs to improve their hard skills — i.e. machine learning techniques and tools, UNIX/Linux, GitHub, editors, etc. — as well as soft skills like corporate communications and time management.
The second kind of candidate is someone with a master’s degree who has been in industry for a few years. That employee will typically know (a lot) about machine learning tools and techniques and can communicate well with product and business managers. This kind of candidate is usually interested in learning more math (especially the “why” from papers and research) and wants to work on skills that they missed out on in prior roles like Spark, Tensorflow/pytorch and general image processing.
What’s the number one trait you look for that you wouldn’t necessarily find on a resume?
There are literally dozens of important things that can’t be estimated from a resume. We screen for all of them, but if a potential employee wants to be proactive I sometimes suggest they maintain a GitHub with a personal project that other people are using (and that’s not just a fork of someone else’s project).
That can be simpler than it sounds — there’s plenty of older code that needs to be converted into python, and there are plenty of data sets (government, public video game, web- or app-collected data and more) that nobody has bothered to analyze yet. That could be you.