MLOps Engineer

About the job About The Role The role is responsible for bridging the gap between machine learning development and production systems by building and maintaining scalable, reliable MLOps infrastructure. This involves designing automated pipelines that take models from training to deployment, ensuring consistency across environments and minimal downtime. The team works closely with data scientists, ML engineers, and infrastructure architects to standardize tools, streamline model delivery, and monitor the health and performance of systems handling large-scale production inference. Key Responsibilities Design and implement automated end-to-end ML pipelines for model training, evaluation, registration, and deployment using tools like Kubeflow or MLflow Deploy, scale, and manage ML inference services on Kubernetes (EKS/GKE) using modern serving frameworks such as Triton Inference Server, Seldon Core, or TorchServe Build robust feature store integrations to serve real-time and batch features consistently across training and serving environments Develop automated monitoring, alerting, and logging systems to detect data drift, concept drift, and latency anomalies in production models Collaborate with infrastructure teams to manage GPU and CPU resource allocation, optimizing compute costs and latency for high-throughput model endpoints What We Are Looking For 3–6 years of experience in MLOps, DevOps, or Software Engineering, with a strong focus on deploying machine learning models to production Proficiency with containerization and orchestration tools, specifically Docker and Kubernetes (including Helm charts and operators) Hands-on experience with ML lifecycle management tools like MLflow, Kubeflow, Prefect, or Airflow Strong Python development skills alongside experience in managing infrastructure as code using Terraform Familiarity with cloud platforms (AWS, GCP, or Azure) and their native machine learning services Bonus: Experience with Triton Inference Server, custom CUDA optimization, or real-time feature engineering using Redis or Feast