Sr. Software Engineer, AI Compiler
TenstorrentJob Title
Sr. Software Engineer, AI Compiler
Role Summary
Work on TT-Forge, Tenstorrent's MLIR-based compiler, to enable efficient execution of AI models on Tenstorrent hardware. The role involves developing compiler passes, custom MLIR dialects, optimizations, and tooling that bridge high-level frameworks to low-level execution.
This is a hybrid role based in Toronto, ON, working closely with hardware and software engineers to produce performant, open-source compiler technology.
Experience Level
Senior. Tenstorrent states candidates at various experience levels may be considered and will be assessed during interview; no explicit years-of-experience requirement is given in the posting.
Responsibilities
Primary responsibilities include building compiler infrastructure and optimizations that map AI models to Tenstorrent hardware.
- Design and implement compiler transformation passes and optimizations in MLIR.
- Develop and maintain custom MLIR dialects (e.g., TTIR, TTNN, TTKernel) and code generation backends.
- Integrate high-level frameworks (PyTorch, JAX, TensorFlow) into the compiler pipeline.
- Collaborate with hardware and systems engineers to target device-specific features and multi-chip execution.
- Improve performance and correctness of model execution, including training and scaling across chips.
- Contribute to tooling for human-in-the-loop optimization and tuning (e.g., TT-Explorer).
- Participate in code reviews, design discussions, and open-source community engagement.
Requirements
Must-have technical skills and practical experience required for the role. Preferred items listed separately.
- Must-have: Fluent in C++ and Python; experience building complex software systems.
- Must-have: Practical experience with compiler optimization or deep systems programming.
- Must-have: Familiarity or strong interest in MLIR and modular compiler frameworks.
- Must-have: Experience working with ML frameworks such as PyTorch, JAX, or TensorFlow, or equivalent experience integrating models to runtimes.
- Nice-to-have: Experience developing custom dialects, transformation passes, or code generators for specialized hardware.
- Nice-to-have: Prior work on multi-chip scaling, runtime systems, or performance tooling.
- Strong collaboration and communication skills; experience working in open-source or cross-functional teams.
Education Requirements
Not specified.
About the Company
Company: Tenstorrent
Headquarters: Austin, Texas, United States
Tenstorrent is a technology company focused on designing innovative computing solutions. They are known for their expertise in the development of advanced hardware, including ASICs and SoCs, aimed at enhancing performance and efficiency in various applications.
