Zheqing (Bill) Zhu is the Founder and CEO of Pokee AI and former head of Applied Reinforcement Learning team at Meta AI. Bill leads Pokee AI to build the next-generation foundation AI agents that excel in reasoning, planning and tool usage. Pokee AI's foundation tool usage model surpasses GPT4o, Claude 3.7 and Gemini 2.5 pro in function calling by a large margin and can easily extend to more than 6000 tools. So far, Pokee AI has secured 12M USD seed round funding from Point72 Ventures, with participation from Qualcomm Ventures, Samsung NEXT, among others as well as top angels such as Lip-bu Tan (CEO of Intel) and Ahbay Parasnis (Founder of Typeface and ex-CTO of Adobe).

Before Pokee AI, Bill was a Senior Staff Research Lead Manager at Meta AI, where he served as the Head of Applied Reinforcement Learning team. He led the development and open-sourcing of Pearl, Meta's flag-ship reinforcement learning training platform for production use cases and deployed RL models across ads, recommender systems, and reality labs, realizing more than 500M USD of annual revenue. Prior to serving as Head of Applied Reinforcement Learning, he was the engineering manager and tech lead for Meta’s Ads Growth Machine Learning team, where he built the first advertiser growth AI product and grew Meta's active advertisers from 2M to 12M.

Bill earned his PhD degree in Reinforcement Learning at Stanford University, advised by Professor Benjamin Van Roy, while working full-time at Meta AI leading the Applied Reinforcement Learning team. His main research focus is to understand theoretical and practical gaps in existing reinforcement learning algorithms in a real-world context. He received Master of Science in Computer Science from Stanford University (also while full-time at Meta AI) and Bachelor of Science in Computer Science with a Minor in Finance, summa cum laude, from Duke University. His work has been highlighted as Meta AI NeurIPS 2023 highlight, CMO highlight launch, multiple win of the month awards at Meta. He has been the recipient of Alex Vasilos Memorial Award, the Highest Distinction Graduate Award from Duke University and Ericsson BUSS Shanghai Quarterly Technical Award. His publications have appeared in top venues including JMLR, ICML, ICLR, KDD, Machine Learning, RecSys, ICRA, IROS and more. 

 Education

• PhD, Reinforcement Learning, Stanford University, Advisor: Benjamin Van Roy, 2023 (Completed while leading the Applied Reinforcement Learning team at Meta AI)

• MS, Computer Science, Stanford University, 2020 (Completed while leading the Applied Reinforcement Learning team at Meta AI)

• BS, Computer Science, summa cum laude, Duke University, Advisor: Ronald Parr, 2017

Honors

• Company-wide NeurIPS AI highlight - Pearl open-source RL library, Meta AI, 2023

• CMO's Highlight Launch List, Facebook (Meta), 2021

• Stanford Graduate Student Fellowship, 2020

• Win of Month / Win of Quarter, Ads Growth, Facebook (Meta), 2017-2021 (Multi-time Winner)

• Alex Vasilos Memorial Award, Duke University, 2017

• Graduate with Highest Distinction, Duke University, 2017

• Ericsson BUSS Shanghai Quarterly Technical Award, 2015

Community Services

• Workshop Chair of AAAI 2023 Reinforcement Learning Ready for Production Workshop

• Workshop Advisor of RLC 2024 Deployable Reinforcement Learning Workshop

• Reviewer: NeurIPS, AAAI, MLJ, ICLR, KDD

Invited Talks

• Reinforcement Learning and AI Agents, World AI Conference (to come), 2025

• Reinforcement Learning and AI Agents, Future Builderz, 2025

• The Year of Breakout for RL Agents, Jinqiu Capital, 2025

• Production-Ready Reinforcement Learning Platform - Pearl, NeurIPS 2023

• Reinforcement Learning for Recommender Systems, Neflix Research, 2023

• Reinforcement Learning for Recommender Systems, DataFunSummit, 2023

• Deep Exploration for Recommendation Systems, at University of Chinese Academy of Sciences, 2023

• Deep Machine Learning Panel, at ML Summit San Francisco, 2019

• Deep Reinforcement Learning Applications, at Shanshu.ai, 2019