About The Role The role bridges the gap between model development and production operations - building the CI/CD pipelines, infrastructure, and monitoring systems that let ML teams ship models quickly and reliably. This is a hands-on MLOps position focused on containerized ML workloads, Kubernetes-based serving infrastructure, and automated training pipelines. The role matters because models that can't be deployed, monitored, and retrained efficiently don't create value. As the AI organization scales, this position will define how experiments move from notebooks to production - and keep them running at a fraction of their current operational cost.

Key Responsibilities

Design and maintain automated ML pipelines for training, validation, and deployment using Kubeflow, MLflow, or Airflow Build and operate model serving infrastructure on Kubernetes (KServe, Seldon, or Triton Inference Server) with autoscaling and GPU optimization Implement CI/CD workflows for ML systems - automated testing, model versioning, staged rollouts, and one-click rollback Establish monitoring and observability for production models: data drift detection, latency/throughput dashboards, and alerting via Prometheus/Grafana Optimize infrastructure costs across cloud platforms (AWS, GCP), including GPU utilization, spot instance strategies, and storage tiering for training data Partner with data scientists to productionize feature pipelines, ensuring training/serving consistency and reproducibility Document architecture decisions and operational runbooks; drive adoption of MLOps best practices across the broader ML organization What We Are Looking For 3–6 years of experience in MLOps, platform engineering, or ML infrastructure, with multiple production model deployments under your belt Strong proficiency with Kubernetes, Docker, and Terraform (or similar IaC tooling) in a cloud environment (AWS, GCP, or Azure) Hands-on experience with ML workflow tooling: MLflow, Kubeflow, Airflow, or Vertex AI Pipelines Solid Python skills and experience with at least one framework for model serving or optimization (FastAPI, TorchServe, ONNX, TensorRT) Practical understanding of ML fundamentals sufficient to debug training pipelines, evaluate model quality, and reason about drift BS in Computer Science, Engineering, or related field (or equivalent practical experience) Bonus: experience with LLM inference infrastructure (vLLM, Triton), distributed training (Ray, DeepSpeed), or building internal ML platforms used by 10+ engineers