TPU/JAX ML Performance Engineer

Pageshift; Zürich; On-site; CHF 130,000 - 190,000; Full time

Job summary

Pageshift is seeking a specialized engineer to profile and optimize long-context JAX workloads on TPUs, focusing on XLA behavior and Pallas kernel development. You will implement and tune kernels, profile memory and sharding, and benchmark changes to validate improvements with data, contributing to the efficiency of large-scale AI storytelling workloads.

You will collaborate with researchers and engineers to iterate on performance optimizations, and bring an experimental mindset, strong

Qualifications

• Experience with JAX, XLA and TPU workloads.
• Ability to profile memory and optimize kernels.
• Experience with benchmarking and data-driven validation.
• Passion for entertainment and storytelling.

Responsibilities

• Implement and optimize JAX workloads for TPUs.
• Develop and tune Pallas kernels.
• Profile memory, sharding and execution bottlenecks.
• Benchmark changes and validate improvements with data.

Skills

JAX workloads

TPUs

Pallas kernels

Profiling

Benchmarking

Kernel optimization

Memory profiling

Job description

Pageshift is seeking a specialized engineer to profile and optimize long-context JAX workloads on TPUs, focusing on XLA behavior and Pallas kernel development. You will implement and tune kernels, profile memory and sharding, and benchmark changes to validate improvements with data, contributing to the efficiency of large-scale AI storytelling workloads.

You will collaborate with researchers and engineers to iterate on performance optimizations, and bring an experimental mindset, strong