SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 100,000 - 500,000 / annual
Tenstorrent is building cutting-edge AI technology with a focus on performance, ease of use, and cost efficiency. The company has developed a high-performance RISC-V CPU from scratch and is creating a unified AI platform spanning software models, compilers, platforms, networking, and semiconductors.
As a Low-Level Software Engineer on the Low-Level Kernel (LLK) team, you will work at the intersection of software and AI hardware to build the foundational primitives that power AI workloads. Low-level kernels sit directly on top of Tenstorrent's AI hardware ISA and form the foundation for the entire software stack above them.
You will tackle challenging problems in low-level programming, hardware-aware optimization, and performance engineering. Your work will directly influence how the hardware is programmed and how efficiently AI workloads execute. Improvements at this layer have outsized impact across the entire stack.
Key responsibilities include:
- Design and implement low-level kernels and software primitives that expose the capabilities of Tenstorrent AI hardware
- Own complex performance optimization efforts, from profiling and identifying bottlenecks through implementation and validation
- Develop software that enables efficient model bring-up and execution across Tenstorrent hardware
- Partner with LLK subteams (performance, tools, compute, architecture bring-up) and collaborate with hardware and runtime teams
- Drive improvements to quality, performance, and capabilities of the low-level software stack
- Take ownership of complex technical problems from design through implementation and optimization
You will learn how to program and optimize software for custom AI hardware at the lowest levels, understand how hardware architecture and instruction sets enable AI workloads, master advanced performance analysis and hardware-aware optimization techniques, and see how low-level decisions influence the broader AI software stack.
The role is hybrid, based in Toronto, Canada. The company welcomes candidates at various experience levels; during the interview process, candidates will be assessed for the appropriate level, and offers will align accordingly.
REQUIREMENTS:
- Strong C/C++ engineer with significant experience building low-level, performance-critical software
- Deep understanding of computer architecture and how software maps to hardware
- Strong problem solver who can investigate complex performance and systems challenges and drive them to resolution
- Comfortable working close to the hardware and making informed tradeoffs between performance, correctness, and maintainability
- Effective technical collaborator who can work across software and hardware teams and influence technical decisions
