Staff Software Engineer, AI/ML Heterogeneous Systems

The Role:
Join the SiFive AI/ML Software Team to build the high-performance stack for Next-Gen AI. We are seeking engineers to optimize and deploy LLMs and Generative AI models on RISC-V architectures, spanning from compiler infrastructure to distributed runtime systems.

Responsibilities:

• Heterogeneous System Architecture: Design and implement the core compute programming model and device runtime daemon. You will define the orchestration of asynchronous task execution, managed memory hierarchies, and efficient host-device synchronization.
• Inference Stack Integration: Lead the deployment of production-grade stacks like vLLM and SGLang on RISC-V. Optimize memory-bound operations, including PagedAttention and sophisticated KV-cache management, within a heterogeneous environment.
• Kernel & Compiler Synergy: Enable seamless lowering from Triton/MLIR-based compiler to SiFive hardware-specific extensions. Develop high-performance kernels that push the limits of RISC-V Vector (RVV) and matrix compute units.
• Co-design: Collaborate with hardware architects to influence the design of future AI accelerators and micro-architectures.