Research Engineer, Infrastructure, Kernels

Thinking Machines Lab is seeking a Research Engineer to join our Infrastructure team, focusing on the development and optimization of machine learning kernels. Our mission is to empower humanity by advancing collaborative general intelligence, and we are dedicated to building AI tools that cater to diverse needs and goals. Our team comprises scientists and engineers who have contributed to widely used AI products and open-source projects.

In this role, you will design and implement custom machine learning kernels using frameworks such as CUDA, CuTe, and Triton, optimizing core operations like attention mechanisms, matrix multiplication, gating, and normalization for modern GPU and accelerator architectures. Collaborating closely with research teams, you will align kernel-level optimizations with model architectures and algorithmic objectives, develop a library of reusable kernels and performance benchmarks, and contribute to the stability and scalability of our infrastructure.

The ideal candidate holds a Bachelor's degree or equivalent experience in computer science, electrical engineering, or related fields, and possesses strong engineering skills with the ability to write performant, maintainable code and debug complex codebases. Proficiency in deep learning frameworks like PyTorch or JAX, as well as GPU programming frameworks such as CUDA, CuTe, or Triton, is essential. Additionally, the candidate should have experience analyzing and optimizing compute-intensive workloads and thrive in a collaborative environment.

This position is based in San Francisco, California. Compensation is competitive and commensurate with experience, and we offer a comprehensive benefits package. Thinking Machines Lab fosters a culture of innovation and collaboration, providing opportunities for professional growth and the chance to work on cutting-edge AI technologies that have a meaningful impact.