Senior Research Scientist AI Theory

Senior Research Scientist – AI Theory

We are seeking a Research Scientist to help develop the theoretical and algorithmic foundations for advanced AI-generated content (AIGC). Our team is engaged in ambitious projects focusing on AIGC theory and the generation of multimodal text-and-image/video content. The ideal candidate should possess expertise in large language models (LLMs), multimodal LLMs, generative modeling (such as diffusion models, autoregressive models, and transformers), and efficient learning techniques (including data efficiency and optimization theory).

The role involves conducting both academic and applied research, with the goal of advancing the state-of-the-art in multimodal modeling. The scientist will contribute to high-impact publications in top-tier conferences and collaborate with other teams to develop cutting-edge products and services.

Key Responsibilities

• Leading and contributing to advanced research projects focused on the theoretical and algorithmic aspects of AIGC.
• Developing, prototyping, and potentially patenting innovative solutions in partnership with product teams.
• Engaging externally through drafting academic publications, delivering talks, and collaborating with academic experts.
• Mentoring research interns and junior researchers.

Person Specification

Required

• PhD in computer science or a related field, or equivalent research experience.
• Practical experience with image/video generation and/or multimodal large language models.
• Proven research track record, including publications in top-tier conferences such as NeurIPS, ICLR, ICML, CVPR, ICCV, and journals like JMLR and TPAMI.
• Strong coding skills with the ability to rapidly prototype in Python.

Desired

• Passion for multimodal large language models and text-to-image/video generation.
• Experience mentoring junior researchers.
• Industry experience and collaboration with product teams.
• Familiarity with tools like PyTorch Profiler, Accelerate/DeepSpeed/Megatron, and HuggingFace PEFT.
• Hands-on experience with ViTs, multimodal LLMs, diffusion models, and (video) VAEs.