Who we look for
We value research drive, technical depth, ownership, and the ability to ask increasingly better questions. Strong candidates do not need to arrive with experience in every DIG direction, but they should be able to learn quickly and push a problem beyond an assigned checklist.
Relevant backgrounds
- Large language models, post-training, RLHF / GRPO, reward modeling, verifiers.
- Agent systems, tool use, long-horizon interaction, memory, skills, and harnesses.
- Machine learning systems, distributed training / inference, RL infrastructure.
- AI for Science, scientific reasoning, experimentation, or domain-specific discovery.
Open roles
PhD students
For students who want to build a coherent multi-year research agenda around post-training, reinforcement learning systems, recursive self-improvement, or discovery intelligence.
Master's students
For research-oriented students who want to develop strong technical foundations and transition toward independent research.
Postdoctoral researchers
For researchers who can own a major direction, mentor junior members, and help shape the group's research identity and infrastructure.
Research interns
Online and on-site research internships are available. Strong interns are expected to work on real research problems and contribute to papers, systems, or open-source artifacts rather than isolated engineering tasks.
How to apply
Email yangling0818@163.com. A concise application is preferred.
Please include
- Your CV or academic resume.
- Your current institution, degree stage, and expected availability.
- Links to representative papers, code, projects, or technical work.
- One short paragraph on the research problems you most want to work on.
- For interns: expected duration and whether you prefer online or on-site work.
You do not need a long cover letter. Clear evidence of what you have built, understood, or discovered is more useful.
Research culture
Problem ownership. The goal is to learn how to define research, not only finish assigned tasks.
High iteration speed. We value fast experimental loops, but speed should increase understanding rather than replace it.
Systems mindset. Papers, code, environments, models, evaluation, and training infrastructure are all part of the research object.
Open impact. We encourage artifacts that other researchers can actually use and extend.
Global collaboration. Strong students will be encouraged to participate in cross-institution research, international exchange, and collaboration with academic and industrial research teams.
Academic & industrial collaboration
DIG welcomes collaboration with academic groups, frontier AI labs, and scientific domain teams on problems aligned with our research agenda. For research collaboration, please contact yangling0818@163.com.
中文
长期招收积极主动、具有较强研究驱动力的博士生、硕士生、博士后和研究实习生，同时欢迎学术界与产业界开展科研合作。欢迎通过邮件联系并附上个人简历、代表性成果以及感兴趣的研究问题。