AI异构硬件性能评估专家 - Seed Model
Core
Build hardware simulation, performance evaluation, and business ROI analysis systems for ByteDance's self-developed LLM training and inference workloads to guide heterogeneous AI accelerator adoption.
Role type
Senior IC AI hardware performance evaluation expert
Builds
Benchmark suites, model profiling, and performance evaluation frameworks for LLM inference and training
Domain
AI infrastructure / HPC / Chip architecture
Deliverable
production ML models
Required skills
Linux C/C++ or Python, computer architecture, chip microarchitecture, high-performance computing, distributed systems, parallel computing, CPU/GPU/NPU architecture, performance bottleneck analysis, ROI analysis
Preferred skills
Hardware simulation, performance modeling, benchmarking, AI accelerator evaluation, Nsight/Profiler/Roofline usage, GPU/NPU software stacks, compiler/operator libraries, capacity planning, TCO analysis
Technologies
Linux, C/C++, Python, Nsight, Profiler, Roofline, LLM frameworks
Responsibilities
Construct benchmarks and model profiles to estimate end-to-end performance of AI accelerators; Analyze constraints between hardware architecture, software stack, and business workloads; Evaluate resource costs, migration risks, and stability to support hardware selection and scaling decisions
Seniority
Senior, hands-on IC
