Experience

I conducted computer vision research advised by Professor Jitendra Malik and Angjoo Kanazawa. I worked on 3D human pose estimation and prediction from video.

I worked in the InterACT Lab under Professor Anca Dragan to develop learning algorithms for high-DOF robots to plan optimal trajectories. I designed methods for robots to correctly interpret and apply corrections made by humans to estimate their cost functions via inverse reinforcement learning. Our paper on Learning from Corrections was presented at ICRA 2019.

I helped develop Statistics 140, better known as Prob 140. The upper-division course, taught by Professor Ani Adhikari, aims to teach introductory probability theory in addition to modern computational tools to simulate, compute, and visualize probabilistic phenomena using Python. My contributions to the course included developing curricula, course materials, class website, course infrastructure, and a Python library.

• ‒ Fall 2018: Head Teaching Assistant for Stat 140: Probability for Data Science. Teaching evaluation: 7.0/7.0 (n=8)
• ‒ Spring 2018: Head TA for Stat 140: Probability for Data Science. Teaching evaluation: 6.6/7.0 (n=11)
• ‒ Fall 2017: TA for Stat 134: Concepts of Probability. Teaching evaluation: 5.9/7.0 (n=9)
• ‒ Spring 2017: TA for Stat 140: Probability for Data Science. Teaching evaluation: 7.0/7.0 (n=1)

I built a web application to automate the process of validating external Python packages and uploading them to a repository for use in internal multi-products. The application significantly reduced the time to perform license and security checks to ensure compatibility with existing products. The backend was built with Flask and MySQL and the frontend was built mostly with jQuery. I also built a data pipeline using Apache Hadoop and Pig to help Python teams monitor post-commit build errors and product health.

I worked on "Smart Recorder," an iOS application written in Objective-C for recording and transcription. I helped move the app from human transcription via Mechanical Turk to machine transcription in addition to a variety of specific features such as Auto-Pause to reduce recording length when input is below a threshold volume.