DevOps (LLM-ops) Freelance

🚀 About the roleWe're looking for a freelancer with hands-on experience running production Kubernetes clusters with GPU support and serving large language models – this isn't a typical DevOps role, but someone who understands the specifics of GPU, CUDA, and LLM inference.

Duration: ~10 working days, engaged on an after-hours basis (not full-time) plus regular calls .

Start potentially in ~3–4 weeks (date not confirmed yet), with possible scope extension.

Tasks

• Deploying and configuring a production Kubernetes cluster with GPU support
• Configuring GPU resource scheduling
• Deploying and configuring LLM serving using vLLM or SG
• Lang
• Configuring CUDA drivers and ensuring compatibility across the GPU stack
• Designing and implementing Infrastructure as Code (Terraform)

Configuring low-latency networking for LLM workloads

• Configuring storage for large model weight files
• Implementing observability (logs, metrics) for the GPU layerMust have
• Experience with Kubernetes in a production environment, including GPU workloads (hosting GPU-based workloads, GPU scheduling)

Experience serving LLM models: vLLM or SG

• Lang
• Understanding of CUDA and GPU drivers
• Experience with Infrastructure as Code (Terraform)

Familiarity with low-latency network configuration

Experience with storage for large model weights

Experience with GPU observability (logs/metrics)

Availability for calls with the client during the projectWe value your trust in providing us your data, thus we are striving to use commercially acceptable means of protecting it. Below you will find the necessary information about the data we collect and process when you send us your CV.

The controller of your personal data is Miquido Spółka z ograniczoną odpowiedzialnością sp. k. with its seat in Krakow, Zabłocie 43a, 30-701 Kraków, Poland (“Miquido”, “We”, “Our” or “Us”). See more