Engineer who lives in the terminal (~13h/day). I focus on the tooling layer that makes AI agents actually useful — Model Context Protocol servers, browser automation that survives bot detection, and the backend services behind them. Technical depth in service of shipping, not ceremony.
- TurboProvider — zero-dependency control panel for llama.cpp that auto-tunes any GGUF for maximum context (MoE expert-offload + quantized KV cache) and serves an OpenAI-compatible endpoint. Runs a 30B MoE coder at 256K context on a 16 GB GPU.
- Skynet Cluster — a local, MCP-native cluster where AI agent harnesses (opencode, Claude Code, …) delegate work and pick it up as sentries: a shared worker registry, task board, message bus, full-text memory, and a live dashboard. Dumb pipes, real agents — they coordinate without knowing anything about each other. Live page ↗
- cc-bridge — read and continue your local Claude Code chats from your phone over LAN/VPN. Zero-dependency Python server + a polished mobile web UI: live updates via SSE, full markdown & tables, and it continues the exact session with full context via
claude --resume. No cloud relay. - Telegram BRO — agent-compatible skill that pings you on Telegram when you're away from the machine. Tuned for 30B+ models.
- Real Browser MCP — exposes a real browser (via puppeteer-real-browser) as MCP tools, so agents can crawl past Cloudflare/Turnstile and behave like a real user.
- Mongo Memory MCP — gives agents persistent memory backed by MongoDB: save, retrieve, and search context across sessions.
- Real Mail MCP — an inbox-management MCP server that actually works, built after too many existing mail servers fell short.
- Scraper MCP — lets agents crawl and scout websites autonomously. WIP, with more on the way.
- Time MCP — time, date, timezone, and scheduling tools for time-aware agent behavior.
- Model Context Protocol servers — building tools that give agents real capabilities
- Browser automation & anti-bot scraping (Puppeteer real-browser, Selenium)
- Local LLMs via Ollama, OCR & intelligent document processing
- Core: Go, Python (FastAPI / Flask), Node.js + TypeScript
- Also: C#, Java, C / C++
- REST, WebSocket, API-key gateways & metered services
- TypeScript, React, Vue, EJS
- HTML + Jinja2, CSS + Tailwind / Bootstrap
- PostgreSQL, MySQL / MariaDB, MS SQL
- MongoDB, Redis, SQLite, Solr
- Docker, Kubernetes, Jenkins, Nginx
- AWS, Azure, Linux, Bash
- Grafana / observability
- 🤖 Agent tooling — expanding the MCP ecosystem with reliable, production-grade servers.
- 🦫 Go backends — fast, metered API services and gateways.
- 🕸️ Resilient automation — scraping and browser automation that holds up against detection.
- LinkedIn · mileamihai@outlook.it
- ☕ Support my work on Ko-fi