Uber’s first head of data science just raised a new venture fund to back nascent AI startups – TechCrunch


Kevin Novak joined Uber as its 21st worker its seventh engineer in 2011, and by 2014, he was the corporate’s head of information science. He talks proudly of that point, however like all good issues, it ran its course and by the tip of 2017, having achieved what he needed on the firm, he left.

At first, he picked up the tempo of his angel investing, work he’d already begun specializing in throughout weekends and evenings, in the end constructing a portfolio of greater than 50 startups (together with the fintech Pipe and the autonomous checkout firm Normal Cognition).

He additionally started advising each startups and enterprise corporations — together with Playground International, Costanoa Ventures, Renegade Companions and Knowledge Collective — and after falling in love with the work, Novak this yr determined to launch his personal enterprise outfit in Menlo Park, Ca., referred to as Rackhouse Enterprise Capital. Certainly, Rackhouse simply closed its debut fund with $15 million, anchored by Uber’s first head of engineering, Curtis Chambers; Steve Gilula, a former chairman of Searchlight Footage, and the fund of funds Cendana Capital. A variety of the VCs Novak is aware of are additionally buyers within the fund.

We caught up with Novak late final week to talk out that new car. We additionally talked about this tenure at Uber, the place, be warned, he performed a significant position in creating surge pricing (although he prefers the time period “dynamic pricing.”) You’ll be able to hear that fuller dialogue or take a look at excerpts from it, edited evenly for size and readability, under.

TC: You have been planning to turn out to be a nuclear physicist. How did you wind up at Uber?

KN: As an undergrad, I used to be finding out physics, math and pc science, and after I bought to grad faculty, I actually needed to show. However I additionally actually appreciated programming and making use of physics ideas within the programming area, and the nuke division had the most important allocation of supercomputer time, in order that ended up driving quite a lot of my analysis  — simply the chance to play on computer systems whereas doing physics. So [I] was finding out to turn out to be a nuclear physicist was funded very not directly via the analysis that ultimately turned the Higgs boson. Because the Higgs bought found, it was superb for humanity and completely horrible for my analysis price range . . .

A pal of mine heard what I used to be doing and kind of knew my ability set and mentioned, like, ‘Hey, you need to come take a look at this Uber cab firm that it’s like a limo firm with an app. There’s a really attention-grabbing information drawback and a really attention-grabbing math drawback.’ So I ended up making use of [though I committed] the cardinal sin of startup functions and wore a go well with and tie to my interview.

TC: You’re from Michigan. I additionally grew up within the Midwest so admire why you would possibly assume that folks would put on a go well with to an interview.

KN: I bought off the elevator and the pal who’d inspired me to use was like, ‘What are you carrying?!’ However I bought requested to hitch nonetheless as a computational algorithms engineer — a title that predated the info science pattern — and I spent the following couple of years residing within the engineering and product world, constructing information options and . . .issues like our ETA engine, mainly predicting how lengthy it will take an Uber to get to you. Considered one of my very first tasks was engaged on tolls and tunnels as a result of determining which tunnel an Uber went via and methods to construct time and distance was a standard failure level. So I spent, like, three days driving the Huge Dig in Boston out to Somerville and again to Logan with a bunch of telephones, amassing GPS information.

I bought to know quite a lot of very random details about Uber cities, however my huge declare to fame was dynamic pricing. . . and it turned out to be a very profitable cornerstone for the technique of creating certain Ubers have been obtainable.

TC: How does that go over, if you inform individuals that you just invented surge pricing?

KN: It’s a really fast litmus take a look at to determine like individuals’s underlying enthusiasm for behavioral econ and finance. The Wall Avenue crowd is like, ‘Oh my god, that’s so cool.’ After which lots of people are like, ‘Oh, thank you, yeah, thanks a lot, great, you purchase the following spherical of drinks’ kind of factor. . . [Laughs.]

However information additionally turned the incubation area for lots of the early particular tasks like Uber pool and quite a lot of the concepts round, okay, how would you construct a dispatching mannequin that allows completely different individuals with pooled experience requests? How do you batch them collectively effectively in area and time in order that we are able to get the suitable match fee that [so this] challenge is worthwhile? We did quite a lot of work on the idea behind the hub-and-spoke Uber Eats supply fashions and considering via how we apply our learnings about ride-share to meals. So I bought the primary individual perspective on quite a lot of these merchandise when it was actually three individuals scribbling on a notepad or riffing on a laptop computer over lunch, [and which] ultimately went on to turn out to be these huge, nationwide companies.

TC: You have been engaged on Uber Freight for the final 9 months of your profession with Uber, so there when this enterprise with Anthony Levandowski was blowing up.

KN: Yeah, it was it was very attention-grabbing period for me as a result of greater than six years in, [I was already developing the] angle of ‘I’ve finished the whole lot I needed to do.’ I joined a 20-person firm and, on the time, we have been closing in on 20,000 individuals . . .and I form of missed the small workforce dynamic and felt like I used to be hitting a pure stopping level. After which Uber’s 2017 occurred and and there was Anthony, there was Susan Fowler, and Travis has this horrific accident in his private life and his head was clearly not within the recreation. However I didn’t need to be the man who was recognized for bailing within the worst quarter of the corporate’s historical past, so I ended up spending the following yr mainly maintaining the band collectively and attempting to determine what I may do to maintain no matter small a part of the corporate I used to be working intact and motivated and empathetic and good in each sense of the phrase.

TC: You left on the finish of that yr and it appears you’ve been very busy since, together with, now, launching this new fund with the backing of outsiders. Why name it Rackhouse? You used the model Jigsaw Enterprise Capital if you have been investing your personal cash.

KN: Yeah. A yr [into angel investing], I had fashioned an LLC, I used to be “marking” my portfolio to market, sending quarterly updates to myself and my accountant and my spouse. It was considered one of these workouts that was a carryover from how I used to be coaching managers, in that I believe you develop most effectively and efficiently for those who can develop a number of abilities at a time. So I used to be attempting to determine what it will take to run my very own again workplace, even when it was simply shifting my cash from my checking account to my “investing account,” and writing my very own portfolio replace.

I used to be actually enthusiastic about the opportunity of launching my first externally dealing with fund with different individuals’s cash beneath the Jigsaw banner, too, however there’s really a fund within the UK [named Jigsaw] and as I began to speak to LPs and was saying ‘Look, I need to do that information fund and I would like it to be early stage,’ I’d get calls from them being like, ‘We simply noticed that Jigsaw did this Collection D in Crowdstrike.’ I spotted I’d be competing with the opposite Jigsaw from a mindshare perspective, so figured earlier than issues go too huge and loopy, I’d create my very own distinct model.

TC: Did you roll any of your angel-backed offers into the brand new fund? I see Rackhouse has 13 portfolio corporations.

KN: There are a number of that I’ve agreed to maneuver ahead and warehouse for the fund, and we’re simply going via the technicalities of doing that proper now.

TC: And the main focus is on machine studying and AI.

KN: That’s proper, and I believe there are superb alternatives outdoors of the normal areas of business focus that, to the extent that you could find like rigorous functions of AI,  are additionally going to be considerably much less aggressive. [Deals] that don’t fall within the strike zone of almost as many [venture] corporations is the sport I need to be taking part in. I really feel like that that chance — no matter sector, no matter geography — biases towards area specialists.

TC: I ponder if that additionally explains the scale of your fund — your wanting to remain out of the strike zone of most enterprise corporations.

KN: I need to guarantee that I construct a fund that allows me to be an energetic participant within the earliest phases of corporations.

Matt Ocko and Zack Bogue [of Data Collective] are good buddies of mine — they’re mentors, in actual fact, and small LPs within the fund and talked with me about how they bought began. However now they’ve a billion-plus [dollars] in property beneath administration, and he individuals I [like to back] are two people who find themselves moonlighting and on the point of make the leap and [firms the size of Data Collective] have mainly priced themselves out of the formation and pre-seed stage, and I like that stage. It’s one thing the place I’ve quite a lot of helpful expertise. I additionally assume it’s the stage the place, for those who come from a spot of area experience, you don’t want 5 quarters of financials to get conviction.



Supply hyperlink