Automotive Human Factors Research

FARLab @ Cornell

Because cars are cool, but people are cooler…

I used to do research with FARLab under Wendy Ju where I worked on a variety of projects helping to build technical solutions and analyzed how people interact both with the technology we create and with each other in various contexts. Some highlighted projects below.

  • Family Car Trip

    A user study analyzing how people interact with each other in a vehicle on long car rides. Understanding moments of conflict and awkwardness and how individuals navigate this and the nuances of a vehicle space that influences these interactions. The vehicle space is special in interactions given close proximity and because everyone is facing the same direction.

  • XR-OOM

    Research around the effectiveness of running in-vehicle user studies with VR. Enables future automotive research with VR interventions for safer, easier research setups. Historically, these research studies did things like having real people walk out in front of a car. Doing this in VR instead is much safer.

    Paper published here.

Prototyping Projects

I get bored sometimes and decide to build random things. Some of these come from work in my free time, other from my time at Cornell Tech as part of my master’s program. Check out these projects below and click the links in the descriptions to learn more.

  • Simon Says Bot

    With posenet, a raspberry pi, pi camera module, two servos, and Google TTS, built a cardboard prototype that could play Simon Says - both as Simon and the follower. More on my Github.

  • Girl Scouts Robotics Mentoring

    Mentored a troop on how to use the Microbit to build an ArtBot and code the bot to draw various shapes.

  • A Day in the Life

    Built out a play on the Game of Life using various Qwiic sensors and a raspberry Pi to make a “game” based on the time of day controlled by the user. More on my Github.

  • Pedometer Pants

    Built a pedometer using various AdaFruit wearable sensors sewn on a pair of pants using conductive thread. Demo. More details.

  • Face and Temperature Detection

    Using a thermal camera and facial landmark detection via OpenCV, dlib, and a raspberry pi (which was very slow). Found the inner canthus and measured the temperature at that point. Video demo.

  • Mirror, Mirror

    Using a proximity sensor, raspberry pi, and servo, build a mirror which could recognize when you stood in front of you and interact with the user. More on my Github.

  • Virtual Hangman Game

    With capacitive sensors, capacitive tape, and a cardboard prototype with letters on hinges, utilized MQTT to create a remote hangman game as the “player” More on my Github.

  • The Bomb Game

    A partially Wizarded game including a series of sensors, raspberry pi, Google TTS, and basic circuitry to make a game for diffusing a fake bomb. More (and bloopers!) on my Github.

  • Linear Optimization of Voxel Dosages

    Used Gurobi to solve a linear optimization problem of determining voxel dosages for different radiation treatment plans based on potential patient needs. More details here.

  • Reducing Racial Bias via Identifying and Removing Latent Variables in Computer Vision Activity Recognition

    Following a paper on mitigating gender bias in imSitu, helped to retrofit this autoencoder model to also mitigate racial bias. More details.

  • BagelBot

    Built out a Slack tool, modeled after Donut, to connect teammates to each other for coffee meet and greets. My co-worker, Ashutosh, who built BagelBot with me, did a write up on the process.

Machine Learning Scientist

AppNexus (Now Xandr)

Helping keep the internet free with online advertising.

In my previous role at AppNexus, I was a data scientist focused on algorithm development for products using a combination of machine learning and heuristics-based techniques. I primarily focused on auction marketplace optimization and big data insights and processing.