I am currently a Quantitative Software Engineer at Two Sigma working on applications of modeling pipelines, software engineering, and data analysis in systematic trading.
Previously, I was a Software Engineer at Cognex, working on deep learning/computer vision applications in factory automation and a researcher at MLCollective, a (very cool!) non-profit machine learning research lab.
I graduated my Master’s in Computer Science at Brown University, working with George Konidaris, Stefanie Tellex, and James Tompkin, in May 2020. Previously, I was an undergraduate student at Brown.
My goal is to create robust, general, and Artificial Intelligence (AI). My interests include Reinforcement Learning (RL), Generalization Across Domains, and Fundamental World Knowledge, with applications to Robots, Games, Finance, Art, and many other domains.
In my free time, I do things like teaching Taekwondo at Columbia, photography, rock climbing, and music.
Publications
Visual Transfer for Reinforcement Learning via Wasserstein Domain Confusion
AAAI 2021 & RL4RealLife 2020, First Author on Paper
Learning Feature Extraction for Transfer from Simulation to Reality
First Author on Undergraduate Honors Thesis at Brown University
Advanced Autonomy on a Low-Cost Educational Platform
Robocup Best Paper Award Finalist, IROS 2019, Second Author on Paper, Poster
PiDrone: An Autonomous Educational Drone using Raspberry Pi and Python
IROS 2018, First Author on Paper, Poster
PiDrone: Design of a Low Cost Autonomous Drone
CARRE International Research Symposium 2017, First Author on Poster
Selected Projects
Generalization in Reinforcement Learning using Metric Learning and Uncertainty Quantification
Current Deep Reinforcement Learning agents overfit to training levels and cannot robustly solve unseen levels of the same task. We learn state abstractions that overcome the generalization gap by learning latent-space metrics and quantifying agent uncertainty. Current project with researcher at McGill & Facebook AI Research.
Learning Representations from Naturalistic Object Transformations in Video
Contrastive algorithms are used for unsupervised learning and improve generalization by learning invariances to specific image transformations present in training data. We apply contrastive learning to sequential (rather than i.i.d.) video frames and learn invariances to naturally occuring transformations, resulting in general object representations. Current project with researcher at MIT & UC Berkeley.
Multi-Recursion Neural RSA
The Rational Speech Acts (RSA) posit that humans act as rational probabilistic speakers and listeners, engaging in recursive reasoning in order to make inferences about the intent and pragmatic meaning behind utterances. We implemented RSA as a neural network and measured the change in accuracy as a function of recursion depth. Project completed at Brown University in 2020.
Awards and Honors
Two Sigma Internal AI/ML Hackathon, Best Project, Two Sigma 2023
NSF Graduate Research Fellowships Program (GRFP) Honorable Mention, National Science Foundation 2020
Senior Prize in Computer Science, Brown University 2019
Michael Black TAship Award, Brown University 2019
Elected to Sigma Xi Honors Society 2019
Academic Honors in Computer Science, Brown University 2019
Robocup Best Paper Finalist, IROS 2019
Research and Teaching Roles
Independent Researcher at Machine Learning Collective - June 2020 Through November 2021
Research Assistant at Brown Robotics, Brown University Department of Computer Science - Summer 2016 through Spring 2020
Graduate Data Scientist at Brown Venture Capital Inclusion Lab - Fall 2019 through Spring 2020
Head Teaching Assistant and Teaching Assistant at Brown CS - Fall 2016 through Spring 2020
Industry Roles
Quantitative Developer (Systematic Macro) at Two Sigma Investments - October 2021 through Present
Machine Learning Algorithm and Software Engineer at Cognex - Summer 2020 through October 2021
Software Engineering and Machine Learning Intern at Cognex - Summer 2019
Deep Learning Intern at NVIDIA - Summer 2018
Systems Programmer, Operator, Consultant (SPOC) at Brown University Department of Computer Science - Fall 2017 through Spring 2019
Talks
2021
Deep Learning: Classics and Trends, MLCollective. *Talk scheduled in May
bigAI, Brown University. To Infinite (Visual) Transfer and Beyond
2020
Nerd Nite Providence, Brown University Love Data Week. Making AI See like Humans
Semi-Finalist at Research Matters Competition, Brown University Graduate School. (Competition/Event cancelled due to COVID-19). Making AI See like Humans
Service
Reviewer & Volunteer for AAAI 2021
Volunteer for Neurips 2020
Reviewer and Program Committee Member for Novel Ideas in Learning-to-Learn through Interaction
Reviewer for Challenges of Real World Reinforcement Learning 2020
Reviewer for Neurips Reproducibility Challenge 2019
Reviewer for International Conference on Robotics and Automation (ICRA) 2020
Reviewer and Technical Program Committee Member for International Symposium on Technology And Society (ISTAS) 2019
Reviewer for IROS 2019
Reviewer for IROS 2018