Principal ML Ops Engineer (EMEA Remote)
Pragmatike · Remote · Ukraine · Czech Republic · Latvia · Spain · Albania · Lithuania · Greece · Bosnia & Herzegovina · Croatia · Estonia · Serbia · Poland · Armenia · Portugal · Italy · Malta · Türkiye · Montenegro · Romania ·
COMPENSATION
€•• – ••k
Listed on the employer's job board (Ashby)
- Fully remote (EMEA)
- Principal level
- Python + Terraform
- GPU infrastructure
- AI-native cloud
Join a well-funded deep-tech startup redefining cloud compute through decentralized, GPU-powered infrastructure. You will design and operate production-grade model serving systems, ensuring high availability and low latency for real-time AI applications while optimizing resource efficiency.
What you'll do
- Build production model serving infrastructure using vLLM, TGI, or Triton
- Implement deployment pipelines with blue/green and canary strategies
- Develop auto-scaling systems and intelligent request routing layers
- Optimize GPU utilization, memory efficiency, and network throughput
- Design observability systems for latency, cost, and system health metrics
- Manage model registries and CI/CD pipelines for automated deployments
- Own full ML system lifecycle including operational support and on-call
What you'll need
- 4+ years in ML Ops, Platform Engineering, or SRE for ML systems
- Hands-on experience with vLLM, TGI, Triton, or equivalent frameworks
- Strong background in container orchestration and GPU workloads
- Experience with MLOps tooling, model registries, and experiment tracking
- Proficiency in Python and infrastructure-as-code tools like Terraform
- Understanding of distributed systems and production reliability engineering
- Ability to use AI coding assistants to accelerate development workflows
- Fluent English required for effective remote collaboration
Nice to have
- Experience with Kubeflow, MLflow, or KubeAI platforms
- Knowledge of GPU scheduling and CUDA/ROCm optimization
- Experience with cost optimization across different GPU types
- Background in early-stage startups or greenfield infrastructure projects
- Proven experience building production systems from scratch
Stack
- vLLM
- TGI
- Triton
- Python
- Terraform
- Helm
- Kubeflow
- MLflow
- CUDA
- ROCm
How you'll work
- Fully remote within EMEA timezone
- On-call responsibilities included
- Start date: ASAP