MLOps Engineer

About The Role

The role owns the infrastructure that makes machine learning work in production — training pipelines, model serving, monitoring, and the CI/CD systems that turn experiments into reliable services.

You will work directly with ML engineers and data scientists to eliminate the gap between a model that works in a notebook and a model that performs reliably at scale, with cost, latency, and reproducibility as first-class requirements.

Key Responsibilities

Build and maintain end-to-end ML pipelines in Kubeflow, Airflow, or MLflow covering data ingestion, training, evaluation, and registration Design and operate model serving infrastructure on Kubernetes (KServe, Seldon, or Ray Serve) with autoscaling, canary deployments, and rollback strategies Implement CI/CD for ML systems using GitHub Actions or GitLab CI, including automated model validation, testing, and staged rollout gates Establish model monitoring for data drift, prediction skew, and latency/cost regression using Prometheus, Grafana, and custom alerting Optimize training and inference costs across GPU fleets — right-sizing instances, batching strategies, quantization where appropriate Manage feature store and data versioning tooling (Feast, DVC) to ensure reproducibility between training and serving environments Document architecture decisions, runbooks, and on-call procedures; share on-call rotation for ML platform reliability What We Are Looking For 3–6 years of experience in MLOps, ML platform engineering, or backend/DevOps engineering supporting ML systems in production Strong Python and bash skills; production experience with Docker, Kubernetes, and at least one workflow orchestrator (Airflow, Kubeflow, Prefect) Hands-on experience deploying and monitoring models at scale on AWS, GCP, or Azure Deep understanding of ML lifecycle tooling: experiment tracking, model registries, lineage, and versioning (MLflow or equivalent) Solid infrastructure-as-code experience with Terraform or Pulumi and GitOps workflows BS/MS in Computer Science, Engineering, or equivalent practical experience Bonus: experience with GPU scheduling (Slurm, Run:ai), LLM inference optimization (vLLM, TensorRT-LLM), or building internal ML platforms used by 10+ engineers