Description du poste
DevOps Engineer
Hybrid, 3 days a week in the office
Hong Kong
Full-timeWe are looking for a DevOps Engineer to own, scale, and secure our cloud and on-premise computing environments. You will work closely with ML researchers, data scientists, and software engineers to streamline model training, optimize GPU cluster utilization, and automate our continuous integration and deployment (CI/CD) pipelines.Core Responsibilities
AI & GPU Infrastructure Management: Provision, monitor, and scale high-performance GPU clusters (NVIDIA H100/A100, CUDA/ROCm setups) for heavy model training and inference workloads.
MLOps & Model Serving Pipelines: Support model deployment, fine-tuning, and low-latency inference optimization using open LLM ecosystems (e.g., vLLM, Triton Inference Server, Llama, Mistral).
CI/CD & Automation: Design, build, and maintain robust CI/CD pipelines for both software services and ML model artifacts using GitHub Actions, GitLab CI, or ArgoCD.
Infrastructure as Code (IaC): Manage cloud resources (AWS, GCP, or Azure) repeatably using tools like Terraform or Pulumi.
Containerization & Orchestration: Deploy, scale, and manage distributed containerized workloads using Kubernetes (K8s), Docker, and KubeFlow or Ray.
Observability & Reliability: Implement logging, tracing, and metric dashboards (Prometheus, Grafana, OpenTelemetry) to track cluster health, GPU memory/utilization, and inference latency.
Security & Governance: Enforce DevSecOps practices, secret management (HashiCorp Vault), container image scanning, and access controls tailored for sensitive AI datasets and models.
Required Qualifications
Experience: 3+ years of experience in DevOps, Site Reliability Engineering (SRE), or Cloud/Platform Engineering, with exposure to MLOps or AI/ML workloads.
Cloud & Containers: Strong hands-on experience with AWS, GCP, or Azure, alongside production Kubernetes and Docker management.
Scripting & Programming: Proficiency in Python, Bash/Shell, or Go for automation and t