At Microsoft AI, we are on a mission to develop the most cutting-edge algorithms for post-training large language models (LLMs) and ship those models to millions of users using Copilot every day. The AI Post-Training team at Microsoft AI is responsible for all aspects of post-training and improving our pre-trained models to advance the state-of-the-art on a wide variety of internal and external benchmarks. Our goal is to push our models’ capabilities in reasoning and instruction following, math, code, and tool use and agentic tasks, among many other areas. This role involves contributions to all stages of the post-training process: driving data collection and acquisition, building evaluations of model capabilities, and applying advanced reward modeling and RL techniques to develop and improve the post-training recipe. We work on the bleeding edge and leverage the most powerful pretrained models and algorithms for our needs. We are an interdisciplinary team of engineers and scientists, learning from each other and collaborating to create the best models. We are looking for outstanding individuals excited about contributing to the next generation of models that will transform the field.

Develop data collection, evaluation, and post-training methods for models.

Design hypotheses and experiment plans for rapidly iterating on model performance.