SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Reality Labs at Meta is seeking a Research Scientist to advance ML hardware-software codesign techniques for next-generation AR, VR, and wearable AI devices. You will work at the intersection of AI models, hardware acceleration, and software systems to develop novel approaches for compute- and power-efficient training and on-device inference of vision and language models.
Key responsibilities include identifying and solving multi-disciplinary ML acceleration problems spanning algorithms, network design, hardware architecture, and multimodal AI use cases. You will collaborate with other research scientists to tackle codesign challenges, inventing novel ML accelerator and system architecture solutions while facilitating integration of algorithms and software to leverage these enhancements.
You will develop state-of-the-art model compression and scalability techniques using numerics, pruning, and distillation. Your work will optimize models on hardware accelerators to meet real-time latency and power constraints, influence partners through data-driven analysis, and define use cases with methodology and benchmarks to evaluate different approaches.
The role requires deep knowledge of how ML acceleration interacts with surrounding systems. You will stay current with latest research advancements, attend conferences, interpret papers, and contribute to patents and peer-reviewed publications.
Required qualifications include a PhD in Electrical Engineering, Computer Science, or relevant technical field (or equivalent practical experience), plus experience developing AI-system infrastructure, AI algorithms, or AI hardware acceleration in C/C++ or Python. Preferred qualifications include experience with PyTorch/TensorFlow, large-scale AI models (CV/LLMs), ML hardware accelerator architecture, performance/power/latency trade-off analysis, and publications at leading venues (NeurIPS, CVPR, ICML, ISCA, MICRO, etc.).
