Backend systems, AI and agent tooling, and infrastructure — most with source on GitHub and a live deployment you can poke at. Most representative work first.
What I build and ship, most of it self-initiated — backend services, AI and agent tooling, and the infrastructure they run on. Source is on GitHub for all of them, and the ones still hosted are linked live.
PHI/PII detection engine that scans code, logs, and API responses for the 18 HIPAA Safe Harbor identifiers — built to catch leaks in developer artifacts before they ship.
LLM pipeline that scores a job posting against your resume and returns an APPLY / CONSIDER / SKIP verdict with evidence — shipping both a deterministic DAG and an agentic planner→executor→synthesizer flow over the same steps.
Browser-based voice agent (Tavus CVI) backed by a FastAPI service that runs seven tool-calling booking functions, persists transcripts, and writes an LLM call summary.
Terraform-managed multi-node k3s cluster on Oracle Cloud's always-free tier — VCN, hardened compute, WireGuard bastion, and remote state. Reproducible from zero.
Upload a PDF, ask questions in natural language, get answers grounded in the source with citations. A FastAPI RAG pipeline over pgvector with JWT auth, self-hosted on Oracle Cloud.
Older projects from while I was learning the stack. Kept for the record.
A collection of fun language tools powered by Gemini AI, with a clean and simple warm-beige UI.
Provisioned and deployed real-time chat application using Terraform scripts on AWS EC2 with DNS routing.
Modular API for managing expenses with JWT-based auth, role-based access, data aggregation, and scheduled summaries.
A computer-vision pipeline for face detection, feature embedding, and similarity-based identity verification, built with TensorFlow and OpenCV.
Chat platform with private messaging, room-based communication, and activity logging using WebSockets.