- Architected agent interfaces for training across multiple simulations and designed and trained deep RL agents using neuro-symbolic methods in complex partially observable simulation environments. - Designed and implemented attention-based entity processing and selection for enhanced deep RL agent perforamnce (in progress)- Lead successful project on formal verification of learning-enabled cyber physical systems funded by the Office of Naval Research under the Small Business Innovation and Research program. (Wrote winning Phase-I and Phase-II proposals, served as technical lead). - Developed probabilistic program in the Gen framework (julia) for tracking entity movements and capabilities, successfully deployed in a military exercise

Studied deep RL agents trained from specialized feedbackPresenting findings at the NeurIPS 2022 ML Safety Workshop

Developed machine learning “Concept Extrapolation” alignment research agendaBuilt technical stack and engineering workflow Lead engineering on variety of ML experimentsWrote and collaborated on posts and internal documents

Won "Great Promise Award", awarded to 3 interns in a class of 15+DARPA Explainable AI (XAI) Program:- Designed and trained causal graphical models to predict deep RL agent behavior - Designed counterfactual explanation algorithm to deliver useful explanationsDARPA Competency Aware Machine Learning (CAML) Program:- Engineering on full stack pipeline in PythonPresented for reinforcement learning reading groupFormulated research directions and conducted literature reviews

Help create the grading rubric for weekly programming assignments.Grade programming assignments, assessing style and efficiency.