Location: Portugal (UTC+0) | Remote only | B2B via US LLC (ACH/Wire)

Technologies: Python, Rust, vLLM, llama.cpp, ONNX, FastAPI, RAG, LoRA fine-tuning, MCP, gRPC

Résumé/Work: https://work.valerii.cc

Email: work[@]valerii.cc

---

AI Systems Architect. I build inference infrastructure, agentic pipelines, and production ML stacks — not prototypes that crumble in prod.

Track record: - Employee #1 at Refact.ai (ex-OpenAI founder). Scraped 80M code repos, co-built dataset for Refact-1.6B-fim (SOTA HumanEval 2022). Built enterprise LLM inference backend, RAG over AST+vectors, SWE-bench-compatible agent loop. - Interim Head of AI (0→1): shipped full air-gapped ML stack in <6 months — training, inference, synthetic data pipelines, fine-tuned BERTs/LoRAs at 70 labels production F1. - Currently: sub-400ms voice-to-voice cascade (STT+VAD+LLM+TTS) on consumer GPU. Zero-dependency MCP server in pure Python.

I operate as Nautiloid Protocol LLC. Clean B2B engagement, no hiring overhead, no equity games.

Looking for: AI infrastructure contracts where output is judged by systems shipped, not meetings attended.