Lead end-to-end ML initiatives: integrate and fine-tune models, optimize GPU inference for latency and throughput, build evaluation and monitoring tooling, collaborate on scalable serving systems, and mentor engineers.