About Neil Traft

Neil Traft on a good hair day
I am currently a member of the Sim ML Team at Aurora Innovation in Pittsburgh. We are building smart actors for self-driving simulations using imitation learning. We believe that there are many behaviors on the road which are inherently cooperative or competitive; to test them, you must have smart, realistic agents who can engage in this game with the AV. Our goal is to generate a diverse repertoire of these smart agents.

Prior to this role switch, I led the Prediction Analytics Team for some years. Prediction is the team responsible for predicting how road users will move, and we served as the "data scientists" of Prediction. In brief, my team: evaluated forecasts despite the unobserved counterfactuals; grappled with problems of representing and evaluating uncertainties; worked to draw correlations between subsystem output and system-level behavior, despite the complex relationship and many confounding factors; explored questions of ML robustness and generalization; all while building an ever-increasing suite of data analysis tools along the way.

Before that, I was a student in the MSc of Computer Science at the University of British Columbia, under Ian Mitchell, studying robotic planning under uncertainty. Before that, I had been a software engineer for six years. Here are some other things about me:

  • For work, I do research and program computers.
  • In my free time, I like to do research and program computers.
  • In another life, I might have been an explorer. I think I would be good at it. But nowadays most of the exploring is already done.
  • In this life, you can usually find me in the office or at the coffee shop, programming computers.
  • To satiate my explorer instinct, I like to travel and go hiking, camping, and backpacking.
  • I have no trouble traveling alone. Sometimes I will program computers while alone.
  • Despite this, I also find pair programming to be a rewarding and enjoyable experience.
  • My main field of interest is computational intelligence, which includes neuroscience, machine learning, robotics, and cognitive science.
  • More generally, I am interested in applying statistics and data science with the goal of furthering knowledge in the natural sciences. Exploring in Red Rock Canyon, Nevada
  • I am deeply interested in the interactions between computation and biology, and how these two fields might inspire and feed off of each other through the broader principles of complexity science.
  • I love thinking about things like pair programming, software engineering methods and practices, and software architecture.
  • My hobbies include programming computers.
  • My hobbies also include running and weightlifting. I believe it's important to stay strong if I am to satisfy my explorer tendencies.
  • Sometimes I yell at computers instead of programming them. Someday in the future, we will program computers primarily by yelling at them.

To connect with me, have a look at the "Being Social" section in the sidebar. And while there's not much there, you can also see some of the code I've written in my GitHub and Bitbucket accounts.

ntraft@gmail.com

Comments are closed.