Moamen Ghareeb

AI Engineer

🎯 voci Real-time speech-to-text and translation overlay for Linux. Streams NVIDIA Parakeet on the local GPU with ~50 ms first-token latency. Optional cloud STT and LLM backends. Python PyTorch NeMo CTranslate2	🎙️ f9-talk Hold F9, speak, release — system-wide dictation that types into the focused window. Single statically-linked Rust binary, distributed as a .deb. Deepgram cloud or offline whisper.cpp. Rust tokio Deepgram whisper.cpp
🛠 What I'm using Building agents with Claude Code, Anthropic SDK, MCP, n8n, OpenRouter, Groq. Inference on PyTorch, NVIDIA NeMo, CTranslate2, ONNX, CUDA. Automation with Playwright, Chrome DevTools, agent-browser, Apify, FastAPI. Infra on Docker, Postgres, Qdrant, Pinecone, Cloudflare Tunnel, Grafana.

---