Uber's First Data Science Head Launches AI Venture Fund

From Nuclear Physics to Venture Capital: Kevin Novak Launches Rackhouse

Kevin Novak began his journey with Uber in 2011 as their seventh engineer. By 2014, he ascended to the position of head of data science for the company. He reflects on this period with pride, acknowledging that all successful endeavors eventually reach their natural conclusion. Consequently, by the close of 2017, having achieved his objectives within the organization, he made the decision to depart.

Angel Investing and the Genesis of Rackhouse Venture Capital

Initially, Novak increased his involvement in angel investing, an activity he had already been pursuing during his evenings and weekends. This led to the creation of a portfolio encompassing over 50 startups, including companies like Pipe in the fintech sector and Standard Cognition, focused on autonomous checkout technology.

Alongside his investing activities, he provided advisory services to both startups and venture capital firms – notably Playground Global, Costanoa Ventures, Renegade Partners, and Data Collective. His passion for this work grew, ultimately prompting him to establish his own venture capital firm, Rackhouse Venture Capital, in Menlo Park, California, this year.

Rackhouse recently finalized its inaugural fund, securing $15 million in commitments. Key investors include Curtis Chambers, Uber’s original head of engineering, Steve Gilula, former chairman of Searchlight Pictures, and Cendana Capital, a fund of funds. A significant number of venture capitalists within Novak’s network also contributed to the fund.

A Conversation with Kevin Novak: Uber, Surge Pricing, and the Future of AI

We spoke with Novak last week to discuss his new venture. The conversation also covered his time at Uber, where he played a pivotal role in the development of what is commonly known as surge pricing – a term he personally prefers to call “dynamic pricing.” Below are excerpts from the discussion, lightly edited for brevity and clarity.

From Physics to Uber: An Unexpected Path

TC: You initially intended to pursue a career in nuclear physics. What led you to Uber?

KN: During my undergraduate studies, I focused on physics, mathematics, and computer science. My goal in graduate school was to become a professor. However, I also enjoyed programming and applying physics principles to computational problems. The nuclear physics department had substantial access to supercomputer resources, which greatly influenced my research – providing opportunities to utilize powerful computers in my work.

My research, indirectly funded through work that ultimately contributed to the discovery of the Higgs boson, faced budget cuts following that breakthrough. A friend alerted me to an opportunity at a “cab company with an app” called Uber, highlighting an intriguing data and mathematical challenge. I applied, despite committing a common mistake for startup applicants by wearing a suit and tie to the interview.

Early Days at Uber: Building the Foundation

TC: Growing up in Michigan, it’s understandable why you might choose a suit for an interview.

KN: Upon arriving, my friend questioned my attire. Nevertheless, I was offered a position as a computational algorithms engineer – a role that predated the widespread use of the “data science” title. I spent the following years working within the engineering and product teams, developing data features and essential components like our ETA engine, which predicted Uber arrival times.

One of my initial projects involved addressing issues with tolls and tunnels, specifically determining which route an Uber took and accurately calculating time and distance. This required extensive data collection, including driving the Big Dig in Boston repeatedly with multiple phones recording GPS data.

I gained detailed knowledge of Uber’s operations in various cities, but my most significant contribution was the development of dynamic pricing, which proved to be a crucial element in ensuring Uber’s availability.

The Legacy of Surge Pricing

TC: How do people typically react when you reveal your role in creating surge pricing?

KN: It serves as a quick gauge of their perspective on behavioral economics and finance. Individuals in the financial sector often express admiration, while others offer more… colorful reactions. [Laughs.]

The data team also served as an incubator for early projects like Uber Pool, exploring how to efficiently dispatch drivers to accommodate pooled ride requests and maintain profitability. We conducted theoretical work on the hub-and-spoke delivery model for Uber Eats, applying insights from ride-sharing to the food delivery sector. I was involved in these products from their inception, often brainstorming with small teams and witnessing their evolution into large-scale businesses.

Navigating a Turbulent Period at Uber

TC: You were involved with Uber Freight during the final nine months of your tenure, coinciding with the controversies surrounding Anthony Levandowski.

KN: It was a challenging period. After more than six years, I felt I had accomplished my goals. I had joined a 20-person company that was nearing 20,000 employees and missed the intimacy of a smaller team. The events of 2017 at Uber – involving Anthony Levandowski, Susan Fowler, and Travis Kalanick’s personal struggles – created a difficult environment.

I didn’t want to be perceived as abandoning the company during a crisis, so I dedicated the following year to maintaining stability and ensuring the team I led remained motivated and ethical.

Rackhouse: A New Beginning

TC: Following your departure at the end of that year, you’ve been remarkably active, culminating in the launch of this new fund with external backing. What inspired the name “Rackhouse”? You previously used “Jigsaw Venture Capital” for your personal investments.

KN: During my initial foray into angel investing, I established an LLC, tracked my portfolio’s performance, and provided quarterly updates to myself, my accountant, and my wife. This was a deliberate effort to develop multiple skills simultaneously, mirroring my approach to training managers – believing that focused development leads to greater success.

I initially considered launching my first externally facing fund under the Jigsaw banner, but discovered another fund in the UK with the same name. As I engaged with potential investors, I realized the existing Jigsaw’s activities might create confusion. To avoid competition for mindshare, I decided to create a distinct brand.

Focus and Strategy for Rackhouse

TC: Did you incorporate any of your angel investments into the new fund? Rackhouse currently has 13 portfolio companies.

KN: I’ve agreed to transfer a few of my angel-backed deals to the fund, and we are currently finalizing the logistical details.

TC: The fund’s focus is on machine learning and AI.

KN: That’s correct. I believe there are significant opportunities outside of traditional industry sectors. Rigorous applications of AI in less competitive areas are particularly attractive. Targeting deals that don’t overlap with the focus of many venture firms is my primary strategy. This approach favors domain experts.

TC: Does the fund’s size reflect this strategy of avoiding competition?

KN: I aim to build a fund that allows me to actively participate in the earliest stages of a company’s development.

Matt Ocko and Zack Bogue of DCVC are mentors and small LPs in the fund. They shared their experiences with me, but their firm now manages over a billion dollars in assets. The companies I prefer to back are typically founded by individuals working on their ventures part-time. Firms like DCVC have priced themselves out of the formation and pre-seed stage, and I find that stage particularly appealing. It’s an area where my domain expertise is highly valuable, and where conviction can be formed without requiring extensive financial data.