Post-training researcher focused on fine-tuning open-source LLMs for specific customer tasks using RL and reward engineering. Involves building training pipelines, environments, and evals, and working with customer data to improve models that reach millions of users.

Research Scientist focused on post-training methodology and performant inference, with a significant portion dedicated to pure research and the remainder to applied research informing the company's platform and customer needs. The role involves designing and executing experiments, publishing at top venues, and collaborating with engineering teams to translate research into production systems.