• Design and operate ML infrastructure: Manage data, training, serving, and inference systems for high-throughput model workflows • Build scalable pipelines: Implement reproducible training and evaluation pipelines with versioning, scheduling, and artifact tracking • Optimize compute and cost: Tune GPU and CPU workloads, manage clusters, and drive efficiency via rightsizing, spot scheduling, and caching • Serve models in production: Operate APIs for low-latency inference with autoscaling, blue-green or canary rollouts, and rollback safety • Ensure reliability and observability: Define and own SLOs; instrument pipelines and services to track latency, cost, drift, and data quality • Secure and automate: Manage IAM, secrets, and container security; automate deployment pipelines via CI/CD and infrastructure as code • Collaborate cross-functionally: Partner with research scientists and AI engineers to deliver models from experiment to production with minimal friction • Document and enable: Build templates, runbooks, and internal tooling that make ML workflows repeatable, safe, and fast