Senior ML Engineer - LLM Training & Serving в Unimatch Lab

Data Scientist / ML Engineer

Вакансия: **Senior ML Engineer - LLM Training & Serving** Компания: Unimatch Lab Формат: #remote Локация: EU, Albania, Andorra, Bosnia and Herzegovina, the United Kingdom, Iceland, Liechtenstein, Moldova, Monaco, Montenegro, North Macedonia, Norway, San Marino, Serbia, Switzerland, Georgia, Armenia, Turkey, LATAM Уровень: Senior Compensation: from $5,000 month base [Unimatch Lab ](https://melted-elm-995.notion.site/Unimatch-Lab-Career-2c39098c69c68034bfdac2e5180e6c02?source=copy_link)is an AI-driven venture studio from Silicon Valley. We build a portfolio of AI products and our own ML stack - including local LLM clusters and in-house training and serving. We are hiring a Senior ML Engineer who owns the full in-house loop: data, fine-tuning, evaluation, and self-hosted production - not prompt-only work or third-party LLM APIs. Personal and sensitive data stay on-premises. [Full job description](https://melted-elm-995.notion.site/Senior-ML-Engineer-LLM-Training-Serving-2bf9098c69c6804ebd5dd73fc0822aa2?source=copy_link) What you'll do: - Fine-tune LLM/VLM models (LoRA/QLoRA, distillation) through production-ready weights - Run eval as a release gate: field-level metrics, held-out sets, train/test leakage checks - Own data workflows: teacher-labeling, versioned datasets, validated synthetic data - Operate self-hosted inference under load: vLLM, batching, quantization, structured JSON outputs - Keep the ML loop documented end to end so it is not tied to one engineer Who we're looking for - 5+ years delivering ML systems to production, including high-load AI - You train model weights yourself - not only prompts or RAG integration - Self-hosted LLM serving in production (vLLM or equivalent) - Full cycle shipped: data -> fine-tune -> eval -> prod traffic - Strong Python, Docker/Kubernetes, CI/CD; in-house ML only for sensitive data If this matches your skills and mindset - apply via https://telegram.me/Unimatch_work #AI #ML #LLM #MLOps #Remote #Hiring