CV

PhD Student

Harvard University

Computer Science

N/A

Bachelor of Arts

Harvard University

Statistics and Mathematics,

Secondary in Computer Science

3.945 / 4.0, magna cum laude.

• Machine Learning
• Stochastic Processes
• Real Analysis

REU Student in Computer Vision and Robotics

Amazon Science Hub, University of Washington

Coordinated with members of UW, NVIDIA, and Amazon Robotics to implement a pipeline for reasoning over bin scenes for online grasping of unknown objects using a robotic arm. Trained and refined models for segmenting unknown objects in bin scenes, generated synthetic data, and applied computer vision techniques to track objects over time.

Mathematics Teaching Assistant

Euler Circle

Assisted in teaching a group of high school students in cryptography and independent project courses. Met with students to learn number theory, probability, and algorithms concepts required for elliptic curve cryptography.

Research Intern in Robotics

Robotics: Perception and Manipulation Lab, University of Minnesota

Worked with Professor Karthik Desingh on pose estimation and manipulation problems for objects with symmetries. Used Blender to create a synthetic dataset and implemented energy-based models to learn orientation distributions.

Course Assistant in Computer Science, Mathematics, and Statistics

Harvard University

Graded problem sets, ran sections and office hours, and tutored students in courses in Linear Algebra, Multivariable Calculus, Reinforcement Learning, and Proofwriting.

Towards Characterizing Curriculum Reinforcement Learning in Sparse Robotics Tasks

Completed a capstone thesis concerning curriculum learning methods in sparse robotic grasping tasks. Recommended for High Honors by a committee of Harvard Professors in the statistics and mathematics departments. Implemented autocurriculum methods in Python using PyTorch and developed visualizations for evaluating success.

• Autocurriculum Reinforcement Learning
• Robotics

Ther-E-Man

Made a theremin that plays sound and controls a custom web game using the hand’s proximity to an antenna. Designed CAD models of the casing using Fusion 360, and printed parts using a laser cutter. Communicated with a HTML website using an ESP-32 web server, and designed the game using JavaScript and Phaser 3.

• Web Design
• Fabrication

Active Inverse Reinforcement Learning for Context-Conditioned MDPs

Developed a framework for actively gathering expert data for learning multi-task IRL policies in a multi-goal maze task. Implemented models for behavior cloning and network-based IRL, and evaluated our results using AL heuristics.

• Pytorch
• Inverse Reinforcement Learning