🤖 Into AI agents, LLM applications, agentic coding, MCP, RAG, and the ecosystem around them.
⚡ I enjoy vibe coding — fast iteration with AI, while still caring about understanding behavior, testing changes, and keeping systems simple.
🔧 Backend-minded. I like turning unclear problems into reproducible bugs and small, verifiable fixes.
I spend most of my time exploring how AI changes the way software is built: coding agents, tool use, context engineering, multi-agent workflows, retrieval, evaluation, and the infrastructure around them.
I also care about the less flashy parts — backend reliability, concurrency, caching, data consistency, search quality, and debugging things until the behavior actually makes sense.
- agentic coding and developer tools
- LLM applications, MCP, tool use, and RAG
- TypeScript and .NET
- CLI / TUI tooling and automation
- backend systems and distributed-system problems
- open-source debugging and small focused fixes
Simple systems. Explicit behavior. Reproducible results.
- #8796 — .NET: Propagate TextSearchProvider caller cancellation
- #2528 — test: run tool activation smoke in CI
- #2527 — fix: support older Git diff prefix options
- #2517 — fix: reject oversized run timeouts
- #698 — fix(jev): keep fallback search within allowed servers
- #697 — fix(config): read UTF-8 BOM-prefixed config files
- #693 — fix(config): preserve malformed config files on write
- #683 → #687 — fix(cache): invalidate stdio metadata on env mode changes
- #678 → #686 — fix(search): dedupe repeated query tokens
