Principal Compiler Engineer
About the team
The compiler team at SambaNova powers the RDU through a unique compilation strategy. By fusing kernels into dataflow graphs that run across the entire accelerator architecture, we avoid sequential kernel launches and eliminate the memory traffic that bottlenecks conventional hardware. This is a key differentiator and unlike any other product on the market.
About the role
As a Principal Compiler Engineer you'll lead critical areas of our compiler including the IR and pass infrastructure, model lowering and partitioning, tensor tiling, memory management, and mapping operations onto the fabric. You'll operate in a tightly coupled environment, influencing hardware design to optimize performance, and bring the technical judgement, cross functional diplomacy, foresight, and low-level programming depth that integrated system design demands. You'll work with senior team members to ensure career growth and engagement, mentor and develop early career talent, actively recruit for the organization, and participate in interviews.
Responsibilities
Lead compiler engineering through ensuring standard methodologies, enterprise product insertion and process evolution
Work with peers, domain experts, developers, customers, and work across the enterprise seeking optimal solutions
Develop, integrate, and implement products
Provide support for proposals in key areas aligned with core team competencies
Basic Qualifications
Bachelor’s or Master’s Degree in Computer Science, Computer Engineering, or equivalent with 5-10 years of industry experience
Additional Qualifications
Deep theoretical understanding of compiler fundamentals
Experience building and deploying software products
Experience with one or more deep learning frameworks (i.e. TensorFlow, PyTorch) is a plus
Experience with common compiler development practices and methodologies
Excitement about high-performance systems engineering and performance debugging
An appreciation for process and developing cross-disciplinary collaboration
Preferred Qualifications
Experience with MLIR
Familiarity with machine learning models and frameworks
Familiarity with accelerated computing
Exposure to dataflow architectures
Base Salary Range:
Base Pay Range
$180,000—$255,000 USD