Senior TPU Inference Systems Engineer

RadixArk; Palo Alto (CA); On-site; USD 200,000 - 400,000; Full time

Job summary

RadixArk is hiring a Member of Technical Staff — TPU Systems to build high-performance inference and training systems using JAX, XLA, and Pallas. You will push model workloads to the limits on TPU hardware and contribute to critical infrastructure enabling frontier models on Google's TPUs.

Responsibilities include optimizing latency, designing SPMD strategies, and developing Pallas kernels. Join a team with deep expertise in AI infrastructure and open-source excellence, including SGLang and

Qualifications

• 3+ years of production ML systems experience.
• Bachelor's or Master's degree in Computer Science, Electrical Engineering, or equivalent.
• Deep understanding of XLA internals (HLO, MLIR, fusion, SPMD, sharding).
• Strong performance tuning across compiler/runtime layers.
• Experience with distributed inference systems or training frameworks.
• Proficiency in Python for high-performance, production-quality code.
• Experience writing custom GPU/TPU/AI accelerator kernels, familiarity with Pallas.

Responsibilities

• Build high-performance inference and training systems using JAX/XLA/Pallas (SGLang-JAX).
• Push large-model workloads to the limits on the latest TPU hardware.
• Optimize end-to-end latency and throughput for LLM serving on TPU infrastructure.
• Design and implement SPMD strategies for distributed inference/training.
• Design and implement Pallas kernels for low-level optimization.
• Profile and optimize XLA compilation and HLO graph transformations.
• Collaborate with kernel engineers and compiler teams for stack-wide performance gains.
• Contribute to open-source TPU optimization guides and benchmarks.

Skills

ML systems experience

Education (BS/MS)

XLA internals

Performance tuning

Distributed inference

Python proficiency

Custom kernels

Education

Bachelor's or Master's degree in CS/EE

Tools

JAX

XLA

Pallas

Miles

Alpa

Pathways

Job description

RadixArk is hiring a Member of Technical Staff — TPU Systems to build high-performance inference and training systems using JAX, XLA, and Pallas. You will push model workloads to the limits on TPU hardware and contribute to critical infrastructure enabling frontier models on Google's TPUs.

Responsibilities include optimizing latency, designing SPMD strategies, and developing Pallas kernels. Join a team with deep expertise in AI infrastructure and open-source excellence, including SGLang and