Bolin Ding

My work is generally about making systems intelligent and efficient with machine learning and optimization techniques. My current research interests include programming frameworks for building LLM agents, systems and algorithms for tuning agent models, efficient data pipelines for training LLMs and agent models, and building novel agent applications (e.g., for data analytics and social science).

Previously, I also worked on database systems, data privacy, data pricing, and federated learning.

I lead several projects including AgentScope (agent programming framework, runtime & infra of agent applications, agent memory, and agent tuning) and Data-Juicer (multimodal data processing for LLMs and agents), which have been widely adopted in production. Some earlier research initiatives include PilotScope (AI4DB middleware) and FederatedScope (federated learning platform).

I'm always interested in discussing research collaborations and speaking opportunities. We are hiring research scientists, engineers, and research interns for our lab!