ML Ops Engineer
On-siteLondon, England, United Kingdom
Job Summary
Provision and manage cloud-native AI/ML infrastructure using Kubernetes, Docker, and GPU orchestration frameworks like NVIDIA GPU Operator or Ray. Automate core platform infrastructure with Terraform, Helm, and Ansible while optimizing GPU compute workloads, high-speed networking, and storage for efficient model training and low-latency inference. Build and maintain robust CI/CD and MLOps pipelines for continuous model training, evaluation, packaging, and production deployment of Large Language Models using advanced inference engines such as Triton Inference Server or vLLM. Enable automated model validation, monitoring for model drift, and end-to-end observability using Prometheus, Grafana, and MLflow. Monitor and optimize cloud spend across high-cost GPU/CPU clusters on AWS, GCP, or Azure by implementing auto-scaling strategies and spot instance policies.
Required Qualifications
- Hands-on production experience in DevOps, Site Reliability Engineering (SRE), or Platform Engineering, with some experience dedicated to AI/ML infrastructure
- Proven track record of deploying, scaling, and operationalising machine learning models and LLMs in cloud-native production environments
- Demonstrated experience managing compute-intensive GPU infrastructure and high-performance computing (HPC) environments
- Advanced proficiency in Kubernetes (K8s), Docker, Helm, KubeFlow, and service meshes (e.g., Istio)
- Hands-on experience with Terraform, Ansible, GitHub Actions, ArgoCD, or Jenkins
- Experience with vLLM, Ray, MLflow, LangChain / LangSmith, DeepSpeed, or Hugging Face TGI
- Solid background in AWS / GCP / Azure, Kubecost, and GPU cost optimisation techniques
- Strong skills in Python, Bash, or Go; deep knowledge of Linux kernel tuning and performance monitoring
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.