MLOps Engineer, LLM Performance

MLOps Engineer, LLM Performance
United States - Remote | Contract | US$90-US$120/hour
HumanitApp is recruiting for an external hiring partner. This is not a position employed directly by HumanitApp. The hiring partner makes selection decisions and confirms the engagement terms.
The role
Use hands-on ML infrastructure experience to create and evaluate technical tasks for advanced AI systems. The work focuses on accelerator kernels, performance profiling, distributed-workload debugging and high-throughput LLM inference. It is an ML systems role, not an applied modeling or general data-science position.
Responsibilities
- Write realistic ML systems tasks and accurate, well-structured solutions.
- Review technical solutions for correctness and provide clear written feedback.
- Assess kernel optimization, profiler traces, distributed systems behavior, and inference throughput and latency trade-offs.
- Develop evaluation criteria and collaborate with other technical specialists.
Required experience
- At least 2 years of hands-on professional experience in ML systems, ML infrastructure, model serving or accelerator performance engineering.
- Production experience with JAX and/or PyTorch.
- Practical depth in at least one area: custom accelerator kernels; performance profiling; distributed/accelerator workload debugging; or LLM serving at scale.
- Experience with modern accelerators and reasoning about compute, memory, latency and throughput.
- Strong technical writing and clear explanations of engineering decisions.
Relevant tools may include CUDA, Triton, Pallas, Nsight, torch.profiler, vLLM, SGLang or TensorRT-LLM. Expertise in every listed tool is not required. JAX, XLA and TPU experience is particularly valuable.
Commitment and terms
- Fully remote; this advert targets candidates working from the United States.
- 40 hours per week during weekdays. This is an exclusive engagement with no conflicting work or other concurrent engagements.
- Advertised compensation: US$90-US$120 per hour; final rate and arrangement are confirmed by the hiring partner. This is not a guaranteed earnings amount.
- Employment or contracting arrangements depend on country and the provider's engagement process. Do not assume W-2 employment or specific benefits.
- The provider currently states it cannot support H1-B or STEM OPT candidates. Final work-authorization requirements are confirmed by the provider.
How to apply
Apply through this LinkedIn listing. You may also email your CV and relevant, non-confidential ML systems experience to [email protected] with the subject "MLOps Engineer, LLM Performance". HumanitApp does not charge an application fee. Please do not send identity documents or confidential employer material.

Work arrangement
Yes