GPU Performance Engineer - Remote, PTO & Visa Support

Megalojobs; San Francisco, Northern (CA, KY); Hybrid; USD 150,000 - 210,000; Full time

Benefits offered by this job

Five weeks of paid leave

Healthcare (vision and dental)

Visa assistance (H1B/OPT)

Job summary

Reka is seeking a Member of Technical Staff (GPU Performance Engineer) to design and optimize training infrastructure for large-scale AI models. You will improve post-training workflows, including reinforcement learning and fine-tuning, and enhance model serving efficiency.

Join an elite team driving performance optimizations in a remote-friendly, globally distributed startup. Ideal candidates will have strong Python/PyTorch skills, CUDA/C++, and experience scaling GPU jobs on large clusters,

Qualifications

• Fluent in Python and PyTorch.
• Proven experience training large deep learning models.
• Experience writing and debugging low-level GPU code (CUDA, C++).
• Experience scaling GPU workloads on large compute clusters (Slurm, Kubernetes).

Responsibilities

• Design and implement improvements to training infrastructure for large-scale models.
• Contribute to post-training processes including reinforcement learning and fine-tuning.
• Improve efficiency and scalability of model serving infrastructure.
• Assist in technical decision-making to optimize model performance.

Skills

Python

PyTorch

CUDA

C++

Slurm

Kubernetes

Profiling

GPU performance

Tools

CUDA toolkit

C++

Slurm

Kubernetes

PyTorch

Python

Job description

Reka is seeking a Member of Technical Staff (GPU Performance Engineer) to design and optimize training infrastructure for large-scale AI models. You will improve post-training workflows, including reinforcement learning and fine-tuning, and enhance model serving efficiency.

Join an elite team driving performance optimizations in a remote-friendly, globally distributed startup. Ideal candidates will have strong Python/PyTorch skills, CUDA/C++, and experience scaling GPU jobs on large clusters,