Projects
Below is a select list of programming projects I've worked on in the past. For a more comprehensive list of my projects, check out my Publications and GitHub.
The RFL Agent
June 2025 - September 2025My main project at Dassault Systèmes SolidWorks. This platform used a locally-hosted AI agent to automate Product Management workflows. These included numerous search-and-summarize tasks, as well as question answering and text summarization tasks. I took responsibility for developing every aspect of this project, including the UI, backend API, and testing framework.
JEBS: A Fine-grained Biomedical Lexical Simplification Task
June 2023 - June 2025The culmination of my two summer research internships at the National Library of Medicine (2023-2024). For this project, I annotated a text simplification dataset of over 1,200 sentence pairs, and developed a suite of language models for identifying and simplifying non-consumer biomedical terms in research abstracts as well as generating explanations for those terms. For this project, I used a number of fine-tuned and off-the-shelf open-source language models.
I presented this work at National Institutes of Health Summer Intern Program Poster Day in both 2023 and 2024, TREC 2024, and ACL 2025. My first-authored paper on this project was published in the Proceedings of ACL 2025.
Grounding Large Language Models with Natural Language World Models
September 2024 - May 2025For my Senior Honors Thesis, I explored how natural language could be used to ground LLMs to embodied environments. To do so, I implemented a novel OpenAI Gym environment for embodied puzzle solving, and compared the performance of LLMs and reinforcement learning agents on various embodied puzzle-solving tasks.
My results reveal a clear divide between what LLMs claim to know and what they actually understand, pointing the way for future work on grounding LLMs to embodied environments.
RAG Pinterest Content Sharing Customer Support Chatbot
April 2025A full stack webapp built around a customer-support LLM chatbot. The chatbot was supplemented with a knowledge graph for retrieval-augmented generation (RAG). I used LangChain and Ollama for the LLM portion of the webapp, with Neo4j serving as the RAG graph database. The backend for this program was held together using Flask and the frontend was implemented with React.js. I created this webapp as part of the interview process for an internship position at Dassault Systèmes.
Alchemist's Dungeon
February 2022A 2D video game I made for my midterm project from when I took CS 23 (Game Design) at Tufts in the spring of 2022. I worked on this project with three other classmates. Alchemist's Dungeon is a single-player puzzle game, where the player must navigate levels with slippery floors to collect potion ingredients.
