I'm a UCLA Computer Science student building AI applications, full-stack software, and embedded telemetry tools. I'm interested in how agents retrieve and update information, and how software works with physical systems.
- Research Intern, UCLA Interconnected & Integrated Bioelectronics Lab (Sep 2026–present): fine-tuning YOLO26 Nano for droplet tracking and building an OpenCV HSV segmentation pipeline, targeting Raspberry Pi deployment.
- Neurosymbolic AI research: developed persistent conversational memory using Python, Scallop, RAG, and Neo4j; co-authored a NeurIPS 2026 LCFM workshop paper.
- AI Engineering Intern, Arvya, Inc. (Jun–Sep 2026): built Outlook/PostgreSQL retrieval and reviewable CRM update proposals for investment banking and private equity teams, with FastAPI and Microsoft Graph.
- UCLA Bruin Racing (Sep 2025–Jun 2026): developed C++ telemetry packets and a Python/pySerial pipeline for decoding sensor data into CSV files; tested malformed packets before hardware integration.
- UCLA Biomimetic Research Lab (Nov 2025–Jun 2026): built simulated EEG, Bayesian inference, and visualization tools using The Virtual Brain, VBI, NumPy, and Matplotlib.
- Liner Notes Search Engine — React/Node.js music search with MusicBrainz and Apple Music, recommendations by genre and release year, and source-cited song comparisons using Tavily and Llama 3.1 on DeepInfra. Tested with Playwright and GitHub Actions.
- RecallAI Meeting Assistant — FastAPI webhook service that schedules meeting bots, generates LLM summaries and Q&A, and emails recaps through Gmail.
- AdmitTrack College Application Tracker — MERN application tracker with authenticated REST APIs and an admissions-data chatbot.
- Memorize Fast and Slow: Symbolic Memory Revision for Test-time Conversational Agents — NeurIPS 2026 Long Context Foundation Models (LCFM) Workshop. Persistent memory that updates facts and handles conflicting information; achieved 67.5% LLM accuracy on a synthetic benchmark. Research code.
- Efficient aspect term extraction using spiking neural network — Neuromorphic Computing and Engineering, 2026.
- Mitigating Bias in Large Language Models: A Collaborative Agent Approach — NeurIPS 2024 Workshop.
Languages: Python, C/C++, JavaScript, TypeScript
AI / ML: PyTorch, OpenCV, Hugging Face, RAG, Anthropic SDK, Scallop
Web / Data: React, Node.js, FastAPI, PostgreSQL, Neo4j, MongoDB
Tools: Git, GitHub Actions, Docker, Azure, Linux/Unix, pytest, Playwright
