Software Engineer, ML Accelerators

DeepMind is building a next-generation compute platform with a strong emphasis on ML hardware acceleration and software and hardware co-design. In this onsite role in Mountain View, you will work on the ML compiler stack and inference infrastructure that connects AI workloads to low-level hardware operators, with a focus on performance tuning and practical debugging.

This full-stack Software (SW) position sits at the intersection of compiler technology and high-performance inference. You will design and develop infrastructure that supports efficient execution, while partnering on SW/HW codesign, kernel optimization, and ongoing performance improvements.

What you will do

• Own end-to-end software development for the ML compiler and inference infrastructure.
• Design and build an ML compiler stack that bridges AI workloads with low-level hardware operators.
• Create software infrastructure that enables high performance inference.
• Drive SW/HW co-design activities, including kernel optimization, performance tuning, and debugging.

What you bring

• Bachelor's degree in Computer Science, a related technical field, or equivalent practical experience.
• Experience with software development using Python and C++.
• Experience with ML hardware accelerators, ML compiler stacks (for example, JAX, PyTorch, and XLA), and kernel development.

Preferred qualifications

• PhD degree in Computer Engineering, Computer Science, or a related field.
• 2 years of experience in full-stack system development for hardware accelerators.
• 2 years of experience in collaborative, hyper-scalar software development and engineering best practices.
• Experience scoping, planning, and executing projects toward team goals in ambiguous environments.

Technologies

• Python
• C++
• JAX
• PyTorch
• XLA

Location: Mountain View, CA (onsite)
Compensation: USD 174,000 - 252,000 per year