Employer description & source details
Own end-to-end performance optimization initiatives for high-impact production ML workloads, from bottleneck identification and implementation through validation and production rollout. Profile and diagnose ML performance bottlenecks across model execution, frameworks, runtimes, memory, CPU/GPU utilization, and infrastructure using rigorous benchmarking and measurement. Design and implement performance improvements involving inference execution, runtime behavior, batching, memory management, precision, compiler integration, custom operators, kernels, or other relevant techniques. Improve ML training and serving efficiency and scalability, optimizing latency, throughput, hardware utilization, and cost while maintaining model quality and production reliability. Support faster and safer model launches through performance benchmarking, load testing, regression detection, and launch-readiness improvements. Partner with model engineers and platform teams to identify high-leverage opportunities, integrate optimizations safely, and validate measurable end-to-end impact. Build reusable performance capabilities, including benchmarks, tooling, optimization patterns, and engineering guidance that accelerate future model development and deployment. Contribute to technical direction and mentor engineers through hands-on collaboration, code reviews, performance debugging, and knowledge sharing. Hands-on ML performance engineering experience: Demonstrated ability to diagnose and implement meaningful performance improvements in production or production-relevant ML workloads. Deep expertise in at least one ML performance area, such as inference execution, ML runtimes, compiler integration, GPU/kernel optimization, memory systems, or training execution. Strong profiling and benchmarking s
Extracted skills: python, machine learning, artificial intelligence