Want to know if this job is worth applying to?
Praha, Česko
Onsite
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.
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?
Coding experience and Infrastructure 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
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.