Sr. Engineer, Software - AI Compiler
TenstorrentJob Title
Sr. Engineer, Software - AI Compiler
Role Summary
Develop and maintain TT-Forge, Tenstorrent's MLIR-based compiler, to map ML models to Tenstorrent hardware. Work on compiler optimization, custom MLIR dialects, transformation passes, and tooling to enable training and inference across single- and multi-chip configurations.
This is a hybrid role based in Santa Clara, CA; Austin, TX; or Toronto, ON, working closely with hardware and software engineers and contributing to open-source components.
Experience Level
Senior-level. The team may consider candidates at adjacent levels during the interview process, but the role expects senior competency in compiler development and system design.
Responsibilities
Key responsibilities include:
- Design and implement compiler optimizations and transformation passes in MLIR.
- Develop and maintain custom MLIR dialects and backend lowering for Tenstorrent hardware.
- Bridge high-level frameworks (PyTorch, JAX, TensorFlow) to low-level execution engines.
- Profile and tune performance across models, hardware configurations, and multi-chip setups.
- Create developer tooling and human-in-the-loop tuning utilities to improve compiler usability.
- Collaborate with hardware and systems engineers and participate in open-source development and reviews.
- Write tests, benchmarks, and documentation for compiler features and optimizations.
Requirements
Must-have and preferred qualifications.
- Must-have: Proficient in C++ and Python; experience building complex systems that bridge frameworks to low-level runtimes.
- Must-have: Practical experience with compiler concepts or strong interest in compiler optimization and MLIR.
- Must-have: Familiarity with ML frameworks (PyTorch, JAX, TensorFlow) or deep systems programming.
- Must-have: Strong collaboration and communication skills for cross-disciplinary engineering work.
- Must-have: Ability to work in a hybrid, team-based environment in one of the listed locations.
- Required for hire: Eligibility to access U.S. export-controlled technology — citizenship, permanent residency, or ability to obtain required licenses may be necessary.
- Nice-to-have: Direct MLIR experience, prior work on custom dialects (TTIR/TTNN/TTKernel-like), experience with multi-chip scaling and compiler tuning tools.
- Nice-to-have: Open-source contributions to compiler or ML tooling projects.
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.
