Zehua Cheng

I obtained by Ph.D. in Computer Science from the Department of Computer Science, University of Oxford.

We build autonomous agents that reason over distributed, high-stakes data (Bio & Finance) without ever centralizing it. We replace trust with cryptography and replace correlation with causal/structural reasoning. For more details, please refer to my WORKING PROJECTS.

Currently, I am leading AI research at FLock.io, where I focus on advancing decentralized ML systems. I am now cooperate with Prof. Jim Davies and Prof Ivan Martinovic from University of Oxford on Decentralized Machine Learning.

News

03/2026 - One paper for SIE has been accepted in ICME 2026! Congratulations to ALL Authors!

03/2026 - I am honored to announce that I will be serving as Vice Chair for the 6th Digital Twin International Conference (DigiTwin 2026). The event will be held in a hybrid format from August 4–8, 2026, at the University of Oxford, UK. DigiTwin is a premier global platform dedicated to advancing digital twin concepts, technologies, and interdisciplinary applications. I look forward to helping organize this prestigious event and collaborating with leading researchers and industry experts. Learn more about the conference here: https://idea-global.net/digitwin2026/index.aspx

12/2025 - I received the Best Paper Award (Industrial) from WI-IAT 2025! Congratulations to ALL Authors!

Open Opportunities:

FLock.io Oxford Phil in Computer Science Scholarship: Security of AI Systems

Selected Awards:

WI-IAT Best Application Award - Scaling Decentralized Learning with FLock
AAAI 2024 Global Competition on Math Problem Solving and Reasoning Track 1 - Solution without API (3rd Prize) and Track 2 - Solution without API (2nd Prize).
Second Prize at KDD Cup 2024 - OAG Challenge AQA.
Third Prize at NeurIPS 2023: Open Catalyst Challenge
International Joint Conference on Artificial Intelligence (IJCAI 2022) Travel Grant
Championship at AWS Self-driving car racing challenge in 2021
Championship at ACM. MM GrandChallenge in Multi-Modal Video Identification Track in 2019