Research Scientist Graduate - (Trust and Safety - CV/NLP/Multimodal LLM, San Jose) - 2026 Start (BS/MS)

Develop state-of-the-art computer vision, NLP, and multimodal models to detect content that violates TikTok’s community guidelines and related regulations. Build and improve content-understanding systems, explore cutting-edge multimodal foundation models, and apply LLM techniques such as pre-training and zero-/few-shot learning. Optimize distributed training frameworks so large models perform reliably at scale.