Principal Engineer, Efficient GenAI
Core
Design and implement efficient Generative AI training and inference solutions at scale for AMD hardware.
Role type
Principal Engineer, Efficient GenAI
Builds
Optimized training and inference pipelines for LLMs, 3D World/Action, and image/video generation models on AMD platforms.
Domain
AI / Deep Learning / High-Performance Computing
Deliverable
production ML models
Required skills
Generative AI model training, distributed training, inference optimization, transformer architectures, parallelism strategies, low-precision training, open-source frameworks (PyTorch, JAX, vLLM, SGLang), algorithmic innovation
Preferred skills
Publications in top-tier conferences (NeurIPS, ICML, etc.), productizing foundation models, agentic workflows
Technologies
PyTorch, JAX, vLLM, SGLang, MuJoCo, AMD hardware
Responsibilities
Propose and apply innovative techniques for training/inference (e.g., speculative decoding, KV-caching), implement novel efficient architectures, collaborate with software/hardware teams for co-optimization, publish research at external venues, promote adoption of agentic workflows
Seniority
Principal, hands-on IC & research
