[Akamai] Performance Engineer

The Akamai Inference Cloud team is part of Akamai's Cloud Technology Group and builds AI platforms for efficient, compliant, and high-performing applications. These platforms support customers in running inference models and empower developers to create advanced AI solutions.

• Experience in performance engineering with hands-on expertise in AI/ML model optimization and inference performance tuning.
• Solid knowledge of inference optimization techniques including quantization (INT8, FP16), model compilation, hardware acceleration, and familiarity with compiler optimizations and ML compilers.
• Proficiency with GPU optimization and understanding of memory hierarchies and techniques to maximize hardware utilization.
• Experience with profiling and benchmarking tools for AI workloads, identifying performance bottlenecks in distributed systems.
• Strong problem-solving skills with ability to analyze performance data, communicate insights clearly, and drive optimization efforts.
• Knowledge of distributed inference and model parallelism techniques.
• Experience with cost optimization for compute-intensive workloads.

• Benchmarking and profiling AI models and inference workloads across different hardware configurations, measuring latency, throughput, and resource utilization.
• Researching and implementing model optimization techniques including quantization, pruning, distillation, and hardware-specific optimizations.
• Optimizing inference frameworks and infrastructure to maximize performance, working with TensorRT, vLLM, TorchServe, Triton and other serving platforms.
• Establishing performance baselines and monitoring for the platform, identifying and addressing performance regressions.
• Collaborating with engineering teams to identify bottlenecks, recommend optimizations, and validate performance improvements.