You'll Be Responsible For:
- Conducting in-depth research into the core technologies of Large Language Models (LLMs) for Function Calling.
- Designing, implementing, tuning, and innovating reinforcement learning algorithms (e.g., PPO, DPO, GRPO) specifically for Function Calling tasks during LLM post-training.
- Leading the design, construction, processing, and analysis of datasets for Function Calling, leveraging data insights to drive model performance.
- Working to improve model performance in practical applications and on public/private benchmarks (e.g., BFCL leaderboard), conducting thorough experimental analysis.
- Documenting innovative findings and experimental results in high-quality technical reports or academic papers.
You Might Thrive In This Role If You:
- Hold a Master's degree or higher in Computer Science, Mathematics, Electronics, or a related field.
- Have 2+ years of experience in LLM algorithms, with a solid foundation in Natural Language Processing (NLP), machine learning, and deep learning, supported by practical experience.
- Are proficient in Python and at least one major machine learning framework, with solid hands-on programming experience in NLP projects; experience in LLM pre-training and fine-tuning is a plus.
- Possess excellent analytical and problem-solving skills, with a passion for tackling challenging problems and a strong curiosity for data exploration.
- Stay current with academic literature and can select and adapt algorithms based on task requirements.
- Are proactive, responsible, and have strong communication and teamwork skills.
Why Join Us?
- Work in a pure research environment, akin to a "West Point for AI."
- Join a team with multiple world championship titles in competitions, with opportunities to contribute to top-tier conference publications.
- Work alongside and grow with scientists recognized in the top 2% of Google Scholar.