Current Work

Technical Lead — Agent Architecture & Engineering Delivery

Jobdream · 2025–Present

Jobdream is a private company founded in 2020 that serves professionals and businesses through AI education and applied agent systems. My work focuses on B2B merchant operations.

• Translate merchant requirements into modular agent workflows, including input parsing, low-stock monitoring, and scheduled freshness checks.
• Test workflows hands-on with the team; use findings to refine system behavior, improve usability for nontechnical owners, and assess readiness for delivery.
• Provide technical oversight for outsourced teams of one to several engineers; evaluate vendors and delivery scope, review architecture and code, and make technical acceptance decisions.

Independent Builder — Agentic Game Platform

Game creation and publishing platform

• Build an agentic platform that lets users create and publish games through vibe coding, coordinating code, 2D animation, shaders, and other assets from concept to playable implementation.
• Use Codex throughout a reviewable feature-delivery loop: translate product intent into scoped changes, inspect architecture and diffs, run automated checks, and iterate against live product behavior.
• Use the platform to develop a LÖVE-based game in Lua, feeding real development friction back into the creation pipeline.

Founder — Investment Education Membership Business

2023–2025 active operation

• Built and operated a $30/month paid membership, reaching approximately 300–400 paying members and an estimated $9,000–$12,000 in peak monthly recurring revenue.
• Grew a YouTube audience to a peak of 9,000+ subscribers and converted viewers into members of the paid Telegram community.
• Deliberately wound down active membership as AI tutoring commoditized the service’s core explanations, encouraging members to build and test their own investment models using the principles and Python-based methods taught in the community.

Reflections on product value and AI

What’s Next

Open to co-founder conversations around the game platform or joining an early-stage team building agentic products.

GitHub · Contact

Research Experience

UCLA — VCLA Summer Research Intern

Center for Vision, Cognition, Learning, and Autonomy · July–September 2019

• Improved a ROS-based glove interface for real-time hand-pose tracking and interaction in the MuJoCo physics engine.
• Developed autoencoder-based hand-pose reconstruction in follow-up collaboration, informing later USENIX Security 2023 research.

USTC — Undergraduate Researcher

MOE–Microsoft Key Laboratory of Multimedia Computing and Communication (MCC) · August 2018–January 2020

• Led reference-frame generation in a four-person team; achieved 28% BD-rate savings and second place in the Water Bottom Video Challenge 2018.
• Applied insights from video coding to later predictive-coding research for federated learning (IEEE JSTSP 2022).

Education

North Carolina State University — Ph.D., Electrical Engineering

2021–2025 · Successfully defended in summer 2025.

Research in machine learning, distributed training, communication efficiency, and privacy.

Ph.D. dissertation: Toward Robust and Secure Federated Learning.

University of Science and Technology of China — B.E., Electronic and Information Engineering

2016–2020

Honorable Title of Excellent Undergraduate Students, USTC — 2020.

University of California, Los Angeles — Summer study

July–September 2019 · Credit-bearing research alongside the VCLA research internship.

Selected Recognition

• Graduate Award, NC State — recurring annual recipient through 2025.
• USENIX Security Student Grant — 2023.
• ICML Participation Grant — 2022.
• UCLA CSST Scholarship — 2019.
• Third place, International Youth Leadership Finance Summit competition — 2019. Team project incorporating AI tools into an online tutoring platform.
• Tang Lixin Scholarship — 2018. Privately funded by entrepreneur Tang Lixin; annual award of RMB 10,000.
• China National Scholarship — 2018. Top 1% distinction.
• Second place, Water Bottom Video Challenge — 2018. Underwater video compression project with HEVC.

Selected Research

My background includes machine learning research at NC State University, with work on distributed learning, communication efficiency, and privacy. Selected publications:

• ICML 2025: NTK-DFL: Enhancing Decentralized Federated Learning in Heterogeneous Settings via Neural Tangent Kernel.
• USENIX Security 2023: Gradient Obfuscation Gives a False Sense of Security in Federated Learning.
• ICML 2022: Neural Tangent Kernel Empowered Federated Learning.
• IEEE JSTSP 2022: Communication-Efficient Federated Learning via Predictive Coding.

Publication list on Google Scholar