Senior Principal Model Bring-up Engineer
Core
Bring new open-weight LLM architectures (dense, MoE, hybrid Mamba/attention) from Hugging Face to production serving on SambaNova's RDU dataflow accelerators.
Role type
Senior Principal Model Bring-up Engineer (ML Systems/Compiler)
Builds
SambaNova's RDU dataflow accelerators and shared inference modules (continuous batching, speculative decoding, FP8/FP4)
Domain
AI Infrastructure / Large Language Models / Dataflow Architecture
Deliverable
production ML models
Required skills
ML systems engineering, compiler engineering, Python/PyTorch internals, C++, transformer inference (KV caching, GQA, MoE, speculative decoding), numerical debugging, large monorepos
Preferred skills
dataflow/spatial compilers, kernel DSLs, FP8/FP4 inference, Mamba/SSMs, expert parallelism, multi-agent AI workflows, open-source contributions
Technologies
MLIR, XLA, TVM, Triton, torch-mlir, vLLM, SGLang, TensorRT-LLM, Bazel, Slurm
Responsibilities
Adapt and compile new LLM architectures for RDU accelerators; build shared inference features like continuous batching and speculative decoding; debug intersections of model, compiler, and hardware; validate numerical correctness against reference models.
Seniority
Senior, hands-on IC
