MLOps Engineer

About the role

The role owns the infrastructure, deployment pipelines, and operational reliability of machine learning systems in production. The focus is on scaling model training, serving high-throughput inference endpoints, and ensuring robust monitoring across cloud environments. The team works closely with machine learning engineers and data scientists to bridge the gap between experimental research and scalable production systems, maintaining high standards for latency, cost-efficiency, and system uptime.

Responsibilities

• Architect and maintain scalable MLOps pipelines using Docker, Kubernetes, and Terraform to automate model training, testing, and deployment
• Design and optimize high-throughput model serving infrastructure on AWS or GCP, ensuring low latency and high availability for production inference
• Implement comprehensive monitoring and observability frameworks for deployed models to detect data drift, concept drift, and performance degradation
• Build automated feature stores and data pipelines to ensure consistency and reusability across training and inference environments
• Establish CI/CD pipelines specifically tailored for machine learning code, model weights, and prompt artifact versioning
• Collaborate with security and engineering teams to enforce governance, compliance, and reproducibility standards for AI/ML systems

Requirements

• 3–6 years of experience in MLOps, DevOps, or machine learning engineering with a heavy focus on production infrastructure
• Strong proficiency in Python, containerization tools (Docker), and orchestration platforms (Kubernetes)
• Hands-on experience with cloud ML platforms (AWS SageMaker, GCP Vertex AI, or Azure ML) and infrastructure-as-code tools (Terraform)
• Deep understanding of CI/CD principles, monitoring stacks (Prometheus, Grafana, Datadog), and distributed data processing frameworks (Spark, Ray)
• Degree in Computer Science, Software Engineering, or equivalent practical industry experience

Skills

• Experience managing large language model inference infrastructure, vLLM, Triton Inference Server, or MLflow (bonus)