TPU Performance Engineer for Cutting-Edge Inference

INFERACT SINGAPORE PTE. LTD.; Singapore; On-site; SGD 167,000 - 335,000; Full time

Benefits offered by this job

Medical coverage

Dental coverage

Vision coverage

Equity

Job summary

Inferact Singapore is seeking a TPU performance engineer to advance vLLM as a premier inference engine on Google TPUs. You will develop and optimize TPU backends, compiler integrations, runtime paths, and benchmarking infrastructure using JAX, XLA, Pallas, and related tooling to achieve frontier inference performance on TPU hardware.

You will work at the boundary of inference systems, kernels, compilers, and hardware architecture, delivering production-ready model serving with clear correctness,

Qualifications

• Bachelor's degree or equivalent in CS, engineering, systems, ML, or similar.
• Hands-on experience building or optimizing TPU workloads with JAX, XLA, Pallas, or related tooling.
• Deep understanding of TPU execution, memory behavior, and performance constraints for ML workloads.

Responsibilities

• Build and optimize TPU backends and related runtime paths.
• Develop benchmarking infrastructure and profiling workflows for TPU workloads.
• Improve production-relevant model serving on TPU with correctness, latency, and throughput targets.

Skills

JAX/XLA/Pallas tooling

TPU performance tuning

Kernel optimization

Performance benchmarking

ML inference systems

Profiling tools

Education

Bachelor's degree or equivalent

Job description

Inferact Singapore is seeking a TPU performance engineer to advance vLLM as a premier inference engine on Google TPUs. You will develop and optimize TPU backends, compiler integrations, runtime paths, and benchmarking infrastructure using JAX, XLA, Pallas, and related tooling to achieve frontier inference performance on TPU hardware.

You will work at the boundary of inference systems, kernels, compilers, and hardware architecture, delivering production-ready model serving with clear correctness,