AI Platform Engineer

About the role

Join Trezor, a leading crypto‑security company, to build and operate a dedicated multi‑GPU AI inference platform hosted in our own datacenter. You will own the end‑to‑end lifecycle of the hardware, operating system, model serving stack, and the internal users who rely on it.

Key responsibilities

• Collaborate with IT to provision, configure, and maintain the GPU server – OS, storage, networking, and observability of utilization, thermals, and latency.
• Deploy, upgrade, benchmark, and tune open‑weight large language models using stacks such as vLLM, SGLang, LiteLLM and Open WebUI.
• Manage quotas, routing, cost reporting and, when needed, integrate closed‑model cloud usage.
• Support developers by maintaining API keys, endpoints, documentation and providing workshops, office hours, and self‑service onboarding guides.

Required profile

• Strong interest in the open‑weight model ecosystem and experience evaluating new model releases.
• Service mindset with a focus on unblocking engineers and writing clear internal guides.
• Fluent English for daily work; Czech is a plus.

Required skills

• Infrastructure as Code: Ansible, Terraform, Docker, Kubernetes.
• Linux systems administration: networking, storage, containers, systemd, kernel‑level troubleshooting.
• NVIDIA GPU operations: drivers, CUDA, NVLink, MIG, nvidia‑smi, DCGM.
• LLM serving tools: vLLM, SGLang, LiteLLM, Open WebUI, llama.cpp, Ollama.
• Model optimisation techniques: fine‑tuning, LoRA, quantization.

What we offer

• High ownership and freedom to build impactful AI infrastructure.
• Option to receive compensation in Bitcoin.
• Flexible working hours, supportive team, and professional development budget.
• Renovated office with gym, massages, games, free parking, and additional benefits such as MultiSport card and yoga classes.