Staff engineer building systems that scale and pipelines that think.
Most of my public work explores one thesis: AI changes the SDLC fundamentally — not as a copilot, but as an autonomous actor with explicit contracts, guard rails, and observable state.
agent-patterns |
Named patterns for multi-agent systems — extracted from real pipelines |
autonomous-dev-loop |
Issue → AI coder → PR → AI reviewer → human merge gate |
ai-dev-tools |
Claude Code toolbox — commands, agents, skills for AI-native dev |
sync-engine |
Hybrid REST/webhook sync — idempotent, failure-tolerant |
fullstack-pilot |
Polyglot multi-service stack — GitOps, ADRs, trade-offs documented |
gitops-homelab |
Bare-metal microk8s, GPU passthrough — GitOps via ArgoCD |
ollama-chat |
Self-hosted LLM chat UI, deployed on the above |