Selected work
Built for production, not for demos.
A sample across production SaaS, applied AI, and enterprise systems — each one built or led hands-on.
FirstGlance
2025–present · CTO · sole engineer
An AI startup-discovery platform that matches investors with emerging startups. As sole engineer I designed, built, and run all of it — a 100-endpoint Flask API, a Next.js web app, Claude-powered pitch-deck extraction, Stripe subscriptions, Firebase auth, and blue-green, zero-downtime deploys on AWS. Hundreds of B2B customers, paid plans live, and a TechCrunch Startup Battlefield 200 pick. I also shipped its iOS & Android apps, since retired as the product went web-only.
FlaskNext.jsClaude APIStripeFirebaseAWSReact Native
firstglance.io — live
ST Engineering iDirect — Satellite NMS
Back-end engineer
Network-management (NMS) software for satellite-communication ground systems, serving operators internationally. Added HEO-satellite support to the map-converter (GXT contour data → GeoJSON) and refactored legacy C++ back-end services into modern Node.js microservices.
C++Node.jsSatellite NMSReal-time
HippoHub — AI Agents
Lead / architect
A multi-agent system for small businesses that drafts press releases and earnings-call scripts and answers questions from a document vault, on OpenAI + Gemini. I led and architected it with my team.
Multi-agentOpenAIGeminiDoc vault
Afiniti — Contact-center systems
Senior engineer
Real-time C++ services behind enterprise contact centers — T-Mobile, UnitedHealth, Verizon: a multi-tenant integration stress-tested at 1,500 calls/minute, live telemetry, and a logging library I made 2× faster.
C++Real-timeDistributedHigh availability
On-device AI tutor
Edge AI
A local-first English tutor on Qwen3 via Ollama. A two-layer design — instant heuristics plus a single LLM call — cut responses from 60–120s to ~5s, with zero data leaving the device.
OllamaQwen3Edge / localPydantic
