Most recently, I have been working in venture capital at DYDX Capital, where I have been finding and evaluating early stage companies using data-driven methods. Conceptually, it is a bit like evaluating talent in sports 🏈⚽️🏀⚾️ . Before that, I was the Director of Data Science at an agtech startup called Hi Fidelity Genetics / Technologies. We were very interested in roots 🌽 !

Below, I walk through a few of the projects I have worked on in the past.

A selection of professional experiences

DYDX Capital

• Demonstrated that demographic factors have predictive power in startup success

  I was able to show that demographic information about founders provides useful predictive information about the likely success of the startup.

• Developed an internal platform to monitor people and companies

  We conceptualized our task as 1) finding companies and 2) prioritizing those companies for follow up. I built two platforms. One to manage historical data and train our models. The other to manage incoming people and company data. A key element of the latter system was to use rules-based, AI-based, and prediction-based methods for narrowing the companies of interest for our team.

Hi Fidelity Genetics / Technologies

• Conceived and co-invented a device, called RootTracker, to measure root systems at scale

  Measuring root growth in the field had historically been a challenge, which limited the use of root characteristics as a target of crop improvement. To overcome this problem, I conceived and co-invented a device that uses capacitance touch sensors to measure root systems at scale. This novel device enabled the optimization of root systems for improved nutrient uptake, stress tolerance, and greenhouse gas reduction.

• Built the RootTracker data platform

  For a single device, RootTracker captured data every 5 minutes. At the height of our work, thousands of devices would stream data during a season. I helped build the system that captured these sensor data and tracked our experiments.

• Developed a novel approach to recapitulating root growth using RootTracker data

  For a species like corn, one can think of a root growing as a random walk. A root starts from where the stem touches the soil and then meanders outwards and down. RootTracker could tell us one point along this random walk, which is limited information as to the appearance of the entire root. Using statistical modeling I overcame this limitation to recreate realistic recapitulations of root growth from these data that captured the temporal and spatial patterns of the underlying ground truth.

• Produced client results

  HFT conducted experiments for itself and others, including ag majors BASF, Bayer, and Corteva. I was responsible for analyzing and communicating the subsequent results. For both internal and external projects, this involved establishing the scientific questions of interest, data exploration and analysis, and then communicating results as a presentation or report. Our most compelling results dealt with greenhouse gas emissions. In particular, our pilot studies showed that variations in root architecture could be used to reduce excess fertilizer-related emissions.

• Built out the team and company

  As employee #1, I helped build the company from the ground up --- inventing the technology, assembling the early team, and then overseeing our data engineering, device engineering, and data science teams as the company grew. Through the writing of grants, I helped raise over $2 million in non-dilutive funding.