AI Accelerator, Software Engineer- Graph Optimization/Compilers

Ampere is a semiconductor design company focused on high-performance, energy-efficient computing for AI, cloud, and edge applications. The Software Engineer will optimize deep learning computational graphs and models for Ampere’s AI accelerator hardware, working across frameworks, compilers, runtimes, kernels, and performance infrastructure.ResponsibilitiesOptimize computational graphs for performance, throughput, latency, memory efficiency, and power efficiency on Ampere AI acceleratorsEnable and optimize models, frameworks, and inference platforms, including PyTorch, Llama.cpp, vLLM, and SGLangDevelop graph-level optimizations such as operator fusion, pattern matching, redundancy elimination, constant folding, layout optimization, memory planning, quantization, and accelerator offloadOptimize transformer and LLM workloads, including dynamic shapes, attention mechanisms, KV-cache management, and mixed-precision executionAnalyze end-to-end performance across frameworks, compilers, runtimes, kernels, and hardwareBuild profiling, benchmarking, validation, and performance-regression infrastructureIdentify bottlenecks using traces, compiler diagnostics, microbenchmarks, and hardware performance dataCollaborate with compiler, runtime, kernel, architecture, hardware, and applications teams on hardware/software co-designContribute to architecture, design reviews, code reviews, documentation, and engineering best practices