AI Platform Engineer

We’re Trezor , a leading company in crypto security that has pioneered the hardware wallet industry as the inventor of the world’s first hardware wallet.

We’re building our own AI inference infrastructure , a dedicated multi-GPU server hosted in our datacenter, so that Trezor employees can use large language models on hardware we control, with no data ever leaving our premises. We are looking for an AI Platform Engineer to own that machine end-to-end: the hardware and OS underneath it with IT’s help, the model serving stack on top of it, and the people who use it every day.

This is a hands-on, broad role . Some days you will be benchmarking a newly released open-weight model, another one sitting with a developer helping them wire an agent into their workflow.

If you like owning a system completely rather than a narrow slice of one, this is for you.

👉 What You’ll Do

Run the machine

Collaborate with IT on owning the full lifecycle of our GPU server: OS, storage, networking…

Set up observability — GPU utilization, thermals, memory, request latency, throughput — and alerting that actually catches problems before users do

Run the models

Bring to life an inference stack (vLLM / SGLang, LiteLLM as a gateway, Open WebUI as the front end as an example)

Deploy, upgrade and tune open-weight models

Evaluate new model releases as they land, benchmark them on our hardware for quality, throughput and latency, and recommend what we should be running

Own quotas, routing and cost/usage reporting across teams

Collaborate in optimizing cloud usage as well, if needed. Sometimes we do have to use closed models

Support the engineers

Be the go-to person for developers integrating the internal models into their tooling — IDE assistants, agents, CI pipelines, internal apps

Maintain API keys, endpoints and documentation; write the internal guides that make onboarding self-service

Run internal enablement: short workshops, office hours, examples of what good usage looks like

Consider how queueing and prioritization will work. Who has priority and what long-term tasks are running over night?

💪 About You

Coding experience and I nfrastructure as Code skills (Ansible, Terraform, Docker/Kubernetes) to automate your own work

Some Linux systems administration : networking, storage, containers, systemd, troubleshooting from the kernel up. IT will collaborate here, though

Genuine interest in the open-weight model ecosystem — you already know which models matter this month

Service mindset : you enjoy unblocking other engineers and writing things down

English for daily work; Czech is a plus

Nice to have

Experience serving LLMs in production or a serious homelab: vLLM, SGLang, llama.cpp, Ollama or similar

Working knowledge of NVIDIA GPU operations: drivers, CUDA, NVLink, MIG, nvidia-smi , DCGM

Datacenter experience: rack power budgets, liquid cooling, hardware vendor support processes

Fine-tuning / LoRA, quantization, model evaluation methodology

🤝 Why Join Trezor

A unique opportunity to be part of a pioneering, security-first company in the crypto industry

A role where you can build, implement, and see the real impact of your work

A high level of ownership and freedom

The chance to work in an open-source company where transparency, trust, and security are part of how we think

Option to get paid in bitcoin

Flexible working hours and a supportive team

Budget for professional development , including training programs, courses, and workshops of your choice

Friendly, open culture with regular company events and fun get-togethers

Renovated offices with a gym, massages, football table, billiards, PlayStation, 3D printer and free on-site parking

Additional benefits such as a MultiSport card, company mobile phone tariff, yoga, fitness classes, and more

👋 Interested? We’d love to hear from you. Send us your CV and a few words about yourself , and we’ll get back to you as soon as we review your application.