Building at the intersection of AI research and engineering.
Specializing in knowledge graphs, GNNs, and high-performance systems.
Engineer by craft.
Researcher by instinct.
I'm a Computer Science student at Wayne State University (B.S. expected May 2027). My work splits between AI research under Dr. Lihui Liu, where I first-authored eICU-KG (a benchmark for predicting ICU medications from clinical data), and building production agentic AI systems on a return offer at Ford Motor Company.
I co-founded Code Emergency, a HIPAA-compliant emergency health platform validated by the Head of Emergency Medicine, and reduced p95 latency by 40% on an $8M ARR government-contracted system at SamSearch.
My focus lives at the seam between deep-learning research and production engineering: knowledge graphs, GNNs, calibration, and selective prediction on the research side; Airflow, Docker, FastAPI, and MCP servers on the systems side. I care about evaluation methodology that actually holds up.
Where I've
made an impact.
Software Engineering Intern · Enterprise Technology
Co-designing the Fleet org's first production agentic AI system with an AI Engineer: deterministic rule layer before model inference, context retrieval across 6 custom MCP servers, LLM reasoning, and post-response validation for auditable line-item decisions. Previously shipped 4 production Airflow ETL pipelines into Snowflake and BigQuery with 25+ schema-validation contracts in CI/CD that block breaking changes before production.
Undergraduate Research Assistant · Machine Learning
First author on eICU-KG, a benchmark for predicting ICU patient medications from clinical data, prepared for a NeurIPS-track workshop. Discovered widely-used benchmarks in this space were leaking answer data into training; built a corrected benchmark and a harder long-tail test set (38% of data). Trained RGCN models in PyTorch with a confidence-based selective-prediction step that lifted accuracy 27.5% on the hardest cases with no loss elsewhere, beating 5 comparable approaches.
Co-Founder & Lead Engineer
Designed, built, and shipped a HIPAA-compliant emergency health platform end-to-end on Django, React, PostgreSQL, and AWS/Vercel. Cut median medical info retrieval time by 20%, validated by the Head of Emergency Medicine.
Software Engineer Intern
Reduced p95 latency by 40% on an $8M ARR production system through async Python processing and database query optimization, informed by live traffic profiling across millions of requests; drove 30% engagement growth.
Selected work.
Let's build
something great.
I'm always open to discussing new opportunities, research collaborations, or ambitious projects. Reach out and let's talk.