I build things that get starred, forked, and merged — not just shipped.
memory-bank has 38 stars and 10 forks — it gives Claude Code persistent, branch-aware memory that cuts wasted tokens by 60-80% per session. Immortal has 21 stars and 3 forks — a self-healing infrastructure engine written in Go that detects failures in milliseconds and auto-recovers with agentic AI.
Outside my own repos: two PRs merged this week into nexu-io/open-design — one fixing a silent asset-bundling bug in inline CSS deploys, one adding a /deploy/preflight API that catches broken references before a single byte is uploaded. Four PRs open in mlflow/mlflow — fixing model logging in air-gapped clusters, auto-tracking git metadata on run start, adding ConversationalCoherence and AgentPlanQuality built-in judges for multi-turn LLM evaluation, and a Claude Code skill for GenAI tracing.
The pattern: find real problems in production systems, then fix them properly.
AI & LLM
Languages
ML & Data
Cloud & Infrastructure
Frameworks
- AWS Cloud Foundations
- AWS Machine Learning Foundations
Ship it. Measure it. Improve it.




