Remote GPU Engineer

Summary

The job is for a GPU Engineer at Reka, a globally distributed AI startup. The role involves improving large-scale training infrastructure and optimizing models' training performance. The ideal candidate has strong engineering skills in Python and PyTorch or other acceleration libraries, experience with low-level GPU code, scaling up GPU jobs via large-scale compute clusters, and proficiency in implementing robust monitoring systems for performance tracking.

Requirements

• Strong engineering skills with fluency in Python and PyTorch or other acceleration libraries
• Experience writing and debugging low-level GPU code (CUDA, Triton) and debugging hardware errors
• Experience scaling up GPU jobs via large-scale compute clusters using Slurm or Kubernetes

Responsibilities

• Design and implement improvements to our large scale training infrastructure
• Help make technical decisions to optimize models' training performance and efficiency

Preferred Qualifications

• Knowledge of advanced filesystems, particularly Ceph, and their integration and optimization in large systems
• Proficient in implementing robust monitoring systems for performance tracking and anomaly detection

Benefits

• 4 weeks paid leave
• Healthcare benefits, including vision and dental
• Visa support (such as H1B and OPT transfer for US Employees)
• Open and inclusive work environment