PKU DLIB (Peking University Deep Learning & Intelligence Lab)
LLM efficient reasoning research under the mentorship of ACL Best Paper authors. Focus areas: native sparse attention mechanisms, test-time training (TTT), and inference-time compute optimization.
Co-founder — AI Robotics Studio
子豪工作室 (Studio with Zihao)
Co-founded an AI robotics content and product studio with 子豪兄 (Tongji, open-source robotics creator & Bilibili UP主). Collaborated on open-source robot hardware projects, produced technical demos and educational content, and explored commercialization of embodied AI applications.
Research Intern — Reinforcement Learning & Robotic Control
Sino-German Industry 4.0 Lab (Tongji × Siemens)
Joint research lab with Siemens. Replaced legacy particle swarm optimization with DQN-based reinforcement learning for robotic arm control. Built 3D.js visualization pipeline; completed SITP project certification.