Meta is seeking an experienced Research Scientist to join their Research & Development teams, specifically focusing on AI Infrastructure. The role is part of two crucial teams: the Kernel team and the E2E Performance team. The Kernel team specializes in maximizing inference performance for Generative AI and Recommendation models through high-performance kernel development, having achieved significant milestones like implementing the first FP8 kernel in Meta's production. The E2E Performance team focuses on optimizing end-to-end performance of AI models using various parallelism strategies and distributed inference techniques.
The ideal candidate will work on cutting-edge problems in AI infrastructure, developing specialized kernels and optimization techniques for large-scale AI models. This position offers the opportunity to work on significant challenges in AI system design and infrastructure efficiency, directly impacting Meta's products and user experiences. The role requires deep expertise in machine learning systems, hardware acceleration, and performance optimization.
The position combines research with practical implementation, requiring both theoretical knowledge and hands-on experience with AI infrastructure and hardware acceleration techniques. You'll be working with state-of-the-art technology in areas like model compression, hardware accelerators, and machine learning compilers. The role offers competitive compensation including base salary, bonus, equity, and comprehensive benefits.
This is an excellent opportunity for someone with a PhD in Computer Science or related field who is passionate about pushing the boundaries of AI infrastructure and system optimization. You'll be part of a team that has already achieved significant breakthroughs, such as enabling AMD GPUs for GenAI production applications, and will continue to drive innovation in the field of AI systems and infrastructure.