Research Intern – Self-evolving Multimodal Foundation Models

Role overview

Develop and prototype efficient methods for next-generation multimodal foundation models, with emphasis on vision systems that can improve and adapt over time. The research spans foundational and applied work, including model optimization and deployment to cloud and edge environments, with opportunities to collaborate across research, engineering, and product teams.

Responsibilities - Explore methods for building capable foundation models with lower cost and improved efficiency. - Research and prototype self-evolving vision models, vision-language systems, unified models, and related approaches. - Investigate automatic model optimization and deployment across cloud and edge settings. - Work with research, engineering, and product partners to turn research ideas into practical solutions, including foundation model, retrieval, and agent-based approaches where suitable. - Contribute to libraries and tools, or publish research findings in leading conferences and journals.

Requirements - Currently pursuing or completing a master’s or PhD in computer science or a related field. - Strong Python and PyTorch coding skills. - Research expertise or publications in compact foundation model development and deployment. - Evidence of research or technical impact, such as influential open-source work or publications at leading AI, machine learning, or computer vision venues. - Self-direction and initiative in proposing and implementing new ideas. - Strong presentation and communication skills for internal and external audiences.

Nice to have

Front-end development experience.

Benefits and work setup

Remote, distributed team across the United States and Canada. The internship pays $50 per hour, with overtime eligibility.