Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will develop post-training methods for code understanding, generation, reasoning, debugging, and validation. You will build training data and interactive environments, research reinforcement learning using execution and tool feedback, create feedback-driven learning methods, and establish evaluation and error-analysis systems for continuous improvement.
Requirements
- Research experience or practical experience in model post-training, reinforcement learning, code intelligence, or related fields
- High-quality publications in international top conferences or journals
- Strong engineering and hands-on ability
- Experience leading or contributing to influential large-scale machine learning projects
- Ability to improve model capabilities through rigorous experiments
- Large-scale distributed training experience is preferred
- Competition awards or high-quality research or open-source contributions are preferred
Responsibilities
- Develop post-training methods for code understanding, generation, reasoning, debugging, and validation
- Build training data and interactive environments for complex code tasks
- Research reinforcement learning methods using execution results, tool calls, and environment feedback
- Develop feedback-driven iterative learning mechanisms
- Establish code-capability evaluation and error-analysis systems