Software Engineer, Acceleration Kernel Development

About the role

You will develop low-level code to maximize the efficiency of machine learning workloads on our custom RISC-V hardware. This position focuses on tuning compute kernels to ensure optimal performance at the instruction and memory levels. We are hiring across multiple seniority levels to join our Toronto-based engineering team.

Key facts

Location: Toronto, Ontario, Canada
Engagement: Hybrid
Team: Acceleration Kernel Development

What you'll do

• Write and optimize compute kernels for parallel machine learning tasks.
• Profile and tune performance to address latency, memory, and bandwidth constraints.
• Collaborate with machine learning engineers to integrate kernel optimizations into production pipelines.
• Maintain and debug the low-level software stack to ensure reliability and speed.
• Work alongside hardware engineers to push the limits of our custom architectures.

Requirements

• Proficiency in C and C++ for building high-performance, efficient code.
• Experience in developing and tuning kernels for parallel or high-performance computing.
• Ability to analyze and improve instruction-level code execution.
• Strong problem-solving skills and a focus on precision in software development.

Skills & tools

• C/C++
• Parallel algorithms
• Performance profiling and debugging
• Machine learning workload optimization
• RISC-V architecture concepts

Practical notes

Compensation ranges from $100,000 to $500,000, including base salary and variable components, determined by experience, education, and location.
Employment is contingent upon your eligibility to access U.S. export-controlled technology. You may be required to provide proof of citizenship or permanent residency, or obtain a license from the U.S. Commerce Department, to comply with U.S.

Export Administration Regulations. Offers may be rescinded if export control requirements cannot be met.

#J-18808-Ljbffr