2 ms·
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, gR
by valaises 9mo ago
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 https://work.valerii.cc
Email: valerii[@]nautiloid.dev
---
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.