I build production-grade AI systems with deep ownership across LLM orchestration, retrieval infrastructure, model serving, and scalable backend architecture. Over the last 5+ years, I have shipped multi-agent platforms, RAG-as-a-Service systems, fine-tuned model serving, and cloud-native backend platforms across pharmaceuticals, energy, SaaS, and developer tooling.
Currently, I work on deep-research agents, healthcare SaaS platforms, event-driven AWS microservices, and evaluation workflows for retrieval quality, hallucination control, and latency-cost tradeoffs. I am most interested in tools that make engineering work faster: local-first context systems, practical CLIs, agent workflows, and AI-native backend infrastructure.
- ContextPool - local-first handoff packets and reusable context packs for AI coding agents.
- pb - GitHub-backed personal pastebin CLI with autosave, cache, image paste, and sync.
- Cedric - backend project generator for Django, FastAPI, and Flask.
- Truth or Bluff - closed-source deep research agent for investigating claims with tool calls.
- ContextPool - reusable context packs for Codex, Claude, Cursor, and MCP-capable tools.
- Truth or Bluff - closed-source claim investigation agent.
- Cubit HQ - closed-source AI financial intelligence platform for founders.
- pb - local-first pastebin CLI backed by GitHub.
- Cedric - Python backend scaffolding for humans and coding agents.
- phonecam-bridge - use a phone camera over Wi-Fi and bridge it into Teams/Zoom through OBS.
- The Real Architecture Behind Production AI Agents
- Building a GitHub Backed Pastebin CLI in Go
- Retrieval Augmented Generation (RAG 101)
- Why We Built (and Killed) Describe AI
- How We Turned Dumb CCTV Footage into a Smart API
