Projects

Production AI, research, and software engineering. Four are full case studies.

Case studies

Natural-language questions become optimized SQL on GPT-4o using schema metadata and RAG, with validation, retry, and cited responses.

PythonOpenAI GPT-4o/o1RAGAWS S3SQL

AI research · Co-author

GameBoyWorlds

An embodied-AI benchmark evaluating frontier and open-source vision-language agents on grounding, navigation, planning, and continual self-improvement.

Co-author · under review at ICLR 2027PythonPyTorchVLMs

Open-source SWE

Simple Mailer

An embeddable npm package for bulk and transactional email with fault-tolerant dispatch, backoff retries, quota handling, and crash recovery.

TypeScriptNode.jsSQLiteNodemailerOAuth2Vitestnpm

LLM fine-tuning

Mini-LLM Puzzle Solver

A 3B-parameter LLaMA fine-tuned on NYT Connections puzzles with hybrid SFT + GRPO, improving perfect-solve rate and cluster purity.

PyTorchHugging Face

More projects

  1. Multi-Agent Monitoring

    Monitor, Diagnosis, and Remediation agents with structured handoffs and human-in-the-loop gates. Python · LangGraph · LLMs

  2. Real-Time Financial Analysis Pipeline

    Financial Q&A bot: streaming and inference pipelines on AWS, news embedded in Qdrant, served through a serverless LangChain REST API with RAG. Python · LangChain · AWS

  3. Stock Portfolio Management App

    REST APIs for buy/sell, holdings, and watchlists; Jenkins CI/CD to AWS ECS with zero-downtime rollouts; a native iOS client. React · Java · Spring Boot 3.3.2

  4. Employee Task Management Dashboard

    Role-based dashboards, task assignment, and deadline tracking. React · Java · Spring Boot