Research Engineer/Scientist (all levels), Efficient Models

Build efficient, scalable generative and multimodal AI models for TikTok’s Vision-Applied Research team. You’ll design and implement methods for large model distillation and compression, including distillation frameworks, model acceleration, and hardware-efficient inference. Work on transferring capabilities from foundation models into smaller models to enable scalable training, optimization, and deployment, with an emphasis on diffusion and autoregressive model efficiency.