SK hynix
Director of AI Workload Intelligence
$250K–$300K
San Jose, CA +1 · Director · Full-time · Posted Sep 1, 2026
- Location
- San Jose, CA
San Jose, California, United States - Visa sponsorship
- New hires must complete Form I-9 verifying identity and legal authorization to work in the United States; the company participates in E-Verify.
About the job
Leads global efforts to assess AI workloads for HBF, define memory data-placement strategies, and guide HBF technology development and product strategy.
What you'll do
- Lead global collaboration on AI workload-based HBF development from research through PoC and productization.
- Analyze AI models and workloads; classify memory objects and define placement, caching, prefetching, and offload criteria.
- Develop HBF-aware benchmarks and provide model-driven input to compiler, runtime, and inference software teams.
What they ask for
- Experience leading global AI system architecture initiatives and cross-functional teams.
- Hands-on experience in AI inference systems, ML systems, LLM serving, model analysis, framework/runtime analysis, or heterogeneous accelerator software.
- Strong understanding of Transformers, LLMs, MoE, recommendation models, embedding/retrieval workloads, and multimodal inference.
- Experience with AI inference stacks and analysis of latency, throughput, memory footprint, bandwidth, and data movement.
Pay
$250,000—$300,000 USD (base salary)
Tools
PyTorch, Hugging Face, vLLM, SGLang, TensorRT-LLM, ONNX Runtime, Triton Inference Server
AI focus
Analyzes LLM and multimodal workloads for HBF applicability, including memory-object classification, data placement, caching, prefetching, offloading, and benchmark development.
Model-written summary · the original posting is authoritative