Building credit infrastructure for small businesses.

Identity and access management, applied to AI systems.

Natural language recommendations for the Books team.

Built TicketAgent, a scoring engine that ranks GitHub issues and recommends bounties.

Built a Raspberry Pi pipeline verifying in-store ad reads via concurrent audio capture and transcription.

Payment integration and verification for a student internship marketplace.

Natural language to CAD pipeline, generating buildable CadQuery models from plain-English briefs.

Selective RAG pipeline and serverless AWS migration for an AI therapy platform.

Medical RAG pipeline over ChromaDB, improving MedQA-USMLE accuracy from 19/50 to 25/50.

PyTorch model predicting 3D heart deformation from pressure and stress, 97.65% training accuracy.

Pose estimation pipeline analyzing children's gaze and movement for behavioral research.

Refactored legacy scripts into a Unity/C# engine, expanding wave variety from 2 to 100+.

Electrochemical CO2 capture research.

Decision Tree and K-Means models predicting Titanic survival; SQL Server ETL and Tableau dashboards.