Senior Machine Learning Engineer

Nebius Token Factory is building an AI training and model post-training capability for frontier model improvement. This role owns the infrastructure that makes large-scale training and RL experiments possible, reliable, reproducible, and efficient. The work sits at the intersection of distributed systems, GPU performance, model training frameworks, RL pipelines, and production engineering. Your responsibilities include designing and running model-training and post-training experiments (including SFT, continued pretraining, DPO/IPO/KTO, RL methods like RLHF/RLAIF, PPO, GRPO), building reward functions and evaluation sets, creating synthetic data and data pipelines, analyzing model-behavior failures, building distributed training and RL infrastructure using Megatron-LM, DeepSpeed, PyTorch FSDP/DTensor, Ray, and more, implementing parallelism strategies, optimizing GPU utilization, and writing experiment plans and runbooks. Must-haves: strong Python and PyTorch engineering, hands-on experience in at least two areas among model training, post-training/RL, data pipelines, or large-scale ML systems, ability to design rigorous experiments, understanding of modern LLM behavior and training bottlenecks, strong communication skills. Nice-to-haves: experience with RL frameworks/pipelines and Megatron-LM/DeepSpeed, CUDA and NCCL familiarity, publications or production impact in LLM domains. In the US, key benefits are Health insurance, 401(k), Parental leave, Remote work reimbursement, Disability & life insurance. Pay: Base Compensation Range $195,200 - $262,200 USD annually. We offer competitive compensation and benefits. This is an onsite role in Palo Alto with US-based benefits.