To further support long-horizon and high-impact AI research, we are launching the Seed Edge Research Initiative. This initiative focuses on developing general intelligence models — models that possess human-like learning abilities, interaction capabilities, and tool-use proficiency. We encourage bold, long-term research with inherent uncertainty and cross-modal, interdisciplinary collaboration. Seed Edge provides a highly flexible research environment and long-cycle evaluation, enabling researchers to pursue transformative AI challenges.
What You Will Work On
By joining the Seed Edge initiative, you will collaborate with us to explore:
1. Explore novel intrinsic reward mechanisms to enable models to engage in autonomous learning and continual self-improvement;
2. Explore the development of long-term memory mechanisms to support the next generation of efficient reasoning models and long-sequence reasoning and modeling.;
3. Advance the boundaries of multimodal perception by developing scalable methods to extract knowledge from massive multimodal data with ultra-low signal-to-noise ratios, while building a solid foundation for multimodal fusion and unified modeling;
4. Study how models can use tools and develop visually grounded action capabilities, including new approaches to foundation modeling for full-modality agents and interactive learning in complex environments.
We also welcome self-initiated research ideas — bring your vision and curiosity to help shape the future of general intelligence.
As part of the application process, we encourage you to prepare a research presentation. We're looking forward to deep technical conversations with you.
Minimum Qualifications
1. Currently pursuing a PhD degree, graduating in 2027 or later, in Computer Science, Artificial Intelligence, Automation, Mathematics, Physics, or related fields;
2. Solid research experience in one or more of the following areas: LLMs, computer vision, multimodal learning, AIGC (AI-Generated Content), or machine learning;
3. Strong curiosity and problem-solving skills, with the ability to independently explore solutions;
4. Strong communication and collaboration skills, with a deep passion for advancing fundamental technologies. Self-driven and open-minded in exploring unconventional ideas through iterative experimentation, aiming to push forward the boundaries of AI research.
Preferred Qualifications
1. Solid understanding of core machine learning principles and algorithmic thinking; prior publications at top-tier conferences/journals (e.g., CVPR, ECCV, ICCV, NeurIPS, ICLR, ICML, SIGGRAPH, SIGGRAPH Asia, etc.);
2. Proficient in C/C++ or Python, with strong programming skills. Recognition in competitions such as ACM/ICPC, NOI/IOI, TopCoder, or Kaggle is highly valued.
3. Experience leading impactful projects in any of the following domains: multimodal learning, LLMs, foundation models, world models, reinforcement learning, or rendering & generative models.
Read Full Description