Uses
The tools behind my day job and side projects. People ask about the home inference server most, so it's first.
Home inference server
- Minisforum AI X1 Pro — AMD Ryzen AI 9 HX 370 with Radeon 890M graphics, 32 GB RAM (planning a 96 GB upgrade to fit 70B models).
- Ollama with Vulkan GPU acceleration —
qwen2.5:14bruns fully on the integrated GPU. - Tailscale so my laptop, phone and side-project servers can reach it from anywhere, without opening ports.
- It powers local transcription, embeddings and chat for my projects, and is one of the providers in Drafting Desk.
Languages & frameworks
- Ruby & Rails — my first love since 2011, and still most of my day job.
- Python for AI work, TypeScript/React for front ends, Scala and Java on the JVM, plenty of SQL.
- Kafka, gRPC, Elasticsearch, PostgreSQL, Kubernetes on the platform side.
AI tooling
- Claude and Codex for coding — with my workflow baked into skills.
- Gemini Live API & Google ADK for voice agents (Rena), LangGraph for multi-agent pipelines, MCP for tool interfaces.
- faster-whisper for transcription and sentence-transformers for embeddings, both self-hosted.
Hosting side projects for ~$0
The full reasoning is in What should you actually use to build your next hobby project?
- Cloudflare Pages for front ends, Fly.io for backends that need persistent state and private networking.
- Neon serverless Postgres and Upstash Redis — both scale to zero.
- GitHub Pages + Jekyll for this site and my static projects (how I set it up).