Member of Engineering (Pre-training and inference fault tolerance)

The Member of Engineering focused on Pre-training and Inference Fault Tolerance at Poolside plays a vital role in enhancing the reliability and fault tolerance of Large Language Models (LLMs) during distributed training. Responsibilities include troubleshooting hardware issues, minimizing GPU idle time, developing recovery tools, and improving checkpointing performance through high-quality coding in Python, C/C++, and CUDA. The position requires a strong background in engineering, a solid understanding of distributed systems and LLM fundamentals, and proficiency with frameworks like Torch. Candidates should possess strong algorithmic skills and be comfortable working within Linux environments.