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.
| Project | Description | Stars | Tech |
|---|---|---|---|
| memory-bank | Persistent memory for Claude Code — 3-5x longer sessions, 60-80% fewer wasted tokens. Branch-aware, self-healing, token-efficient. | Python | |
| Immortal | Self-healing engine for modern infrastructure — detects failures in milliseconds, heals autonomously with agentic AI. 79 Go packages, single binary. | Go | |
| global-emergency-map-ai | AI-powered real-time global emergency mapping and intelligence system. | Python |
AI & LLM
Languages
ML & Data
Cloud & Infrastructure
Frameworks
- AWS Cloud Foundations
- AWS Machine Learning Foundations
Ship it. Measure it. Improve it.

