Research Engineer, Infrastructure, Kernels

Thinking Machines Lab•San Francisco, CA

13h•Onsite

About The Position

Thinking Machines is hiring an infrastructure research engineer to design, optimize, and maintain the compute foundations for large-scale language model training. The role develops high-performance machine-learning kernels, enables efficient low-precision arithmetic, and improves the distributed compute stack used to train large models. It involves working close to hardware and across the research boundary, collaborating with researchers and systems architects to connect algorithmic design with hardware efficiency. The engineer will prototype kernel implementations, profile performance across hardware generations, and help define numerical and parallelism strategies for scaling next-generation AI systems. This is an evergreen role: applications are reviewed on an ongoing basis, and an immediate opening may not always match an applicant’s experience. Applicants are encouraged to apply and may reapply after gaining more experience, but should not apply more than once every six months. Separate postings may also be available for specific projects or teams. The position is based in San Francisco, California. The expected annual salary range is $400,000–$500,000, depending on background, skills, and experience. Thinking Machines sponsors visas and offers generous health, dental, and vision benefits, unlimited PTO, paid parental leave, and relocation support as needed.

Requirements

Bachelor’s degree or equivalent experience in computer science, electrical engineering, statistics, machine learning, physics, robotics, or a similar field.
Strong engineering skills, including the ability to contribute performant, maintainable code and debug in complex codebases.
Understanding of deep learning frameworks such as PyTorch or JAX and their underlying system architectures.
Ability to thrive in a highly collaborative environment with cross-functional partners and subject-matter experts.
Initiative and a bias for action, including willingness to work across stacks and teams to help deliver work.
Proficiency in CUDA, CuTe, Triton, or other GPU programming frameworks.
Demonstrated ability to analyze, profile, and optimize compute-intensive workloads.

Nice To Haves

Experience training or supporting large-scale language models with tens of billions of parameters or more.
Track record of improving research productivity through infrastructure design or process improvements.
Experience developing or tuning kernels for deep learning frameworks such as PyTorch or JAX, or for custom accelerators.
Familiarity with tensor parallelism, pipeline parallelism, or distributed data processing frameworks.
Experience implementing low-precision formats such as FP8, INT8, or block floating point, or contributing to related compiler stacks such as XLA or TVM.
Contributions to open-source GPU, ML systems, or compiler optimization projects.
Prior research or engineering experience in numerical optimization, communication-efficient training, or scalable AI infrastructure.

Responsibilities

Design and implement custom ML kernels, such as CUDA, CuTe, and Triton kernels, for core LLM operations including attention, matrix multiplication, gating, and normalization, optimized for modern GPU and accelerator architectures.
Design compute primitives to reduce memory-bandwidth bottlenecks and improve kernel compute efficiency.
Collaborate with research teams to align kernel-level optimizations with model architecture and algorithmic goals.
Develop and maintain reusable kernel libraries and performance benchmarks that support internal model training.
Contribute to infrastructure stability and scalability, reproducibility across precision formats, and high utilization of compute resources.
Document and share insights through internal talks, technical papers, or open-source contributions.