M
ML Engineer — Pre-Training
full-time
senior
About this role
Work on Mistral's next-generation model pre-training pipeline. Design training recipes, optimize distributed training, and push model capabilities forward.
Paris-based with top-tier compute. Competitive with the best labs globally.
Requirements
Strong ML fundamentals. Experience with distributed training (FSDP, DeepSpeed, Megatron). Python/PyTorch required. Publications preferred.