Requirements

Expected technologies

Kubernetes

PyTorch

Terraform

Helm

Kustomize

Optional technologies

AWS

Operating system

Windows

Our requirements

• 5+ years of experience in MLOps, infrastructure, or platform engineering.
• Experience setting up and scaling training and fine-tuning pipelines for ML models in production environments.
• Strong expertise in Kubernetes, container orchestration, and cloud-native architecture (AWS preferred), specifically with GPUs.
• Hands-on with training frameworks like PyTorch Lightning, Hugging Face Accelerate, or DeepSpeed.
• Proficiency in infrastructure-as-code (Terraform, Helm, Kustomize) and cloud platforms (AWS preferred).
• Familiar with artifact tracking, experiment management, and model registries (e.g., MLflow, W&B, SageMaker Experiments).
• Strong Python engineering skills and experience debugging ML workflows at scale.
• Experience deploying and scaling inference workloads using modern ML frameworks.
• Deep understanding of CI/CD systems and their role in ML production.
• Working knowledge of monitoring and alerting systems for ML workloads.
• A strong sense of ownership and commitment to quality, security, and operational excellence.

Your responsibilities

• Design and maintain infrastructure for training, evaluating, and deploying machine learning models at scale.
• Manage GPU orchestration on Kubernetes (EKS), including node autoscaling, bin-packing, taints/tolerations, and cost-aware scheduling strategies (e.g., spot/preemptible GPUs).
• Build and optimize CI/CD pipelines for ML code, data versioning, and model artifacts using tools like GitHub Actions, Argo Workflows, and Terraform.
• Manage and optimize containerized ML workloads on Kubernetes (EKS), including node auto-scaling, GPU orchestration, and runtime scheduling.
• Develop and maintain observability for model and pipeline health (e.g., using Prometheus, Grafana, OpenTelemetry).
• Collaborate with Data Scientists and ML Engineers to productionize notebooks, pipelines, and models.
• Implement and work with security and compliance to bring best practices around model serving and data access
• Support inference backends including vLLM, Hugging Face, NVIDIA Triton, and other runtime engines and Optimize GPU utilization
• Develop tools to simplify model deployment, rollback, and A/B testing for experimentation and reliability.
• Lead incident response and debugging of performance issues in production AI systems.