Member of Technical Staff, Model Training (post-training research)

What You'll Do7

• Research, implement, and scale advanced post-training methods such as supervised fine-tuning (SFT), reinforcement learning (RL), and on-policy distillation (OPD) across large language and multimodal models.
• Probe learning objectives, optimization dynamics, exploration strategies, and credit assignment mechanisms to improve training stability, sample efficiency, and capability acquisition.
• Develop and refine training recipes by jointly tuning data mixtures, sampling strategies, curricula, and evaluation methodologies.
• Design methods for learning from expert demonstrations, human preferences, corrections, and model-generated trajectories, pinpointing where human feedback yields the greatest gains.
• Study the co-evolution of generators, reward models, and graders, creating approaches that lift both model performance and the quality of training signals.
• Design rigorous experiments, ablation studies, and evaluation frameworks to measure generalization, reasoning, and performance on realistic downstream tasks.
• Own research initiatives end-to-end, from spotting promising directions and implementing new algorithms to running large-scale experiments and validating results.

Requirements7

• Solid foundations in machine learning, deep learning, optimization, and modern model training techniques.
• Hands-on experience training, fine-tuning, or post-training large-scale language or multimodal models.
• Deep technical expertise in at least one area such as reinforcement learning, model distillation, reward modeling, data selection, or learning dynamics.
• Strong proficiency in Python and PyTorch, with the ability to implement, debug, and optimize complex training pipelines.
• Experience designing controlled experiments, interpreting model behavior, and turning empirical findings into actionable research directions.
• Ability to independently navigate ambiguous research problems, formulate hypotheses, and iterate based on experimental evidence.
• Strong analytical thinking and a research-driven mindset, with emphasis on reproducibility, scientific rigor, and measurable improvements.

Who Should Apply

This role suits researchers or engineers with hands-on experience post-training large language or multimodal models and deep expertise in reinforcement learning, distillation, or reward modeling. You should be comfortable with Python and PyTorch, designing controlled experiments, and driving research from hypothesis to validated result.

Salary Insight