RadixArk
Member of Technical Staff — Inference-TPU
$200K–$400K
Palo Alto +1 · On-site · Staff · Full-time · Posted Dec 30, 2025
- Location
- Palo Alto
San Francisco - Experience
- 3+ years.
- Education
- Bachelor's or Master's degree in Computer Science or Electrical Engineering, or equivalent industry experience.
About the job
Build and optimize high-performance inference and training systems for large models on TPU hardware.
What you'll do
- Develop inference and training systems, including SGLang-JAX, and optimize LLM serving latency and throughput.
- Design distributed SPMD strategies and Pallas kernels; optimize XLA compilation and HLO transformations.
- Collaborate across kernel and compiler teams and contribute TPU optimization guides, benchmarks, and architectural insights.
What they ask for
- 3+ years of experience building production ML systems using JAX/Torch, XLA, or TPU-focused frameworks.
- Bachelor's or Master's degree in Computer Science or Electrical Engineering, or equivalent industry experience.
- Strong performance-tuning skills across compiler and runtime layers.
- Experience with distributed inference or training frameworks, proficiency in Python, and experience writing custom GPU/TPU/AI Accelerator kernels.
Nice to have
- Deep understanding of XLA internals, including HLO, MLIR, operator fusion, SPMD partitioning, and sharding strategies.
- Familiarity with Pallas for kernel development.
Pay
$200,000 - $400,000 USD (annual base salary)
Tools
JAX, Torch, XLA, TPU, SGLang-JAX, Pallas, Python, HLO
AI focus
Build and optimize inference and training infrastructure for large language models on TPU hardware.
Model-written summary · the original posting is authoritative