Model Research, Optimization, and Training

About Tenstorrent

Tenstorrent is building advanced AI technology that is reshaping expectations around performance, usability, and cost. The company’s work spans software models, compilers, platforms, networking, and semiconductors, with a team that has created a high-performance RISC-V CPU from the ground up. Tenstorrent values teamwork, curiosity, and solving difficult technical challenges, and is open to people at different seniority levels.

Role Overview

The ML Models team works where machine learning research meets high-performance hardware. This group brings modern ML models onto Tenstorrent’s custom AI accelerators, covering everything from large language model training to inference optimization at scale. The position is aimed at candidates who want to help move cutting-edge research into practical, production-ready AI systems.

Work Location

This is a hybrid position based in Toronto, Ontario, with also mention of Boston, Massachusetts as an associated location.

Candidate Level

Applicants from a range of experience backgrounds are welcome. Tenstorrent will evaluate candidates during the interview process and place them at the appropriate level, which may differ from the title shown here.

Responsibilities

• Drive research and development work centered on LLM training and inference performance.
• Train, benchmark, and tune state-of-the-art AI models on Tenstorrent hardware.
• Boost model performance using methods such as speculative decoding, quantization, kernel fusion, flash attention, and distributed training.
• Analyze performance constraints and work with cross-functional teams to improve throughput and efficiency.
• Convert advanced ML research into scalable solutions that are ready for production use.

Requirements

• Strong working knowledge of Python and PyTorch for building and training deep learning models.
• Solid understanding of ML architectures, LLM training workflows, and inference optimization techniques.
• Practical experience training large-scale machine learning systems.
• At least 4 years of experience in machine learning research and/or LLM development in industry or academia.
• A PhD, published research, or experience with speculative decoding is considered a strong advantage.

What You Will Learn

• How to tune AI models on custom accelerators from the application layer down to silicon.
• How large-scale ML systems are deployed, optimized, and scaled in production environments.
• How hardware, compiler, kernel, and ML teams coordinate to improve performance.
• The practical tradeoffs involved in scaling modern AI workloads across custom hardware.

Compensation and Benefits

Total compensation for engineers at Tenstorrent is stated to range from $100k to $500k, including base pay and variable compensation targets. The actual offer depends on experience, skills, education, background, and location. Tenstorrent also indicates that it provides a competitive compensation package and benefits.

Eligibility and Legal Notice

Employment is subject to the applicant being eligible to access U.S. export-controlled technology. Because of U.S. export laws, including the Export Administration Regulations (EAR), compliance checks may be required for individuals from certain countries or those in certain residency categories. This applies to people in the U.S. as well as outside the U.S. If the role cannot be filled due to export-control restrictions, any employment offer may be withdrawn.