Research Engineer, Frontier Speculative Decoding

As a Research Engineer, Frontier Speculative Decoding, you will bridge cutting-edge research and production deployment. You will emphasize data-centric development, meticulous hyperparameter tuning, and rigorous checkpoint evaluation before models hit production. The role involves tailoring general-purpose models to customer-specific data recipes and proprietary datasets to deliver highly efficient, specialized tools. You will design and iterate speculator algorithms to push accuracy–efficiency tradeoffs while leveraging dedicated GPU clusters. Collaborate with customers and cross-functional teams to translate needs into production-ready solutions for an inference platform. You will own technical challenges end-to-end with a focus on reliability, performance, and scalability in a fast-paced environment.