About
I am a researcher at the Agent Systems Center, Shanghai Artificial Intelligence Laboratory, where I build scientific-research agent platforms — constructing the scientific tool graph, generating and screening high-quality agent trajectories, and performing agent post-training.
I received my B.Eng. (2023) and M.Eng. (2026) degrees in Biomedical Engineering from Sun Yat-sen University, and spent 2024–2025 as a research assistant in the AI Thrust of the Information Hub at The Hong Kong University of Science and Technology (Guangzhou).
My research lies at the intersection of AI and healthcare: generative modeling of physiological signals (ECG panorama synthesis and single-lead → 12-lead reconstruction), unobtrusive vital-sign monitoring, and — more recently — AI agents for science.
News
- 07/2026Joined the Agent Systems Center, Shanghai AI Laboratory.
- 2026One first-author paper, Nef-Net v2, accepted to ICLR 2026.
- 11/2025Released Nef-Net+ on arXiv; code is publicly available.
- 2025–26Healthcare research intern at CICC and Zhonggeng Fund; AI engineer intern at Huawei.
- 2024First-author paper published in Biomedical Signal Processing and Control (cited 20+).
Publications
(* equal contribution, bold = myself)
-
Nef-Net v2: Adapting Electrocardio Panorama in the WildInternational Conference on Learning Representations (ICLR), 2026.
-
On the Panoramic Electrocardiogram of Artificial Intelligence: Ideas, Methods, and ApplicationsJournal of Electrocardiology, 2025.
-
Conditional generative adversarial network driven variable-duration single-lead to 12-lead electrocardiogram reconstructionBiomedical Signal Processing and Control, vol. 95, 106377, 2024.
-
Anatomy-embedded Multimodal Segmentation Emulating Real-World Medical Imaging ProcessIEEE Transactions on Medical Imaging, under review.
-
Edge2Analysis: A Novel AIoT Platform for Atrial Fibrillation Recognition and DetectionIEEE Journal of Biomedical and Health Informatics, 26(12): 5772–5782, 2022.
Experience
Building a scientific-research agent platform: scientific tool-graph construction, generation and quality screening of effective agent-trajectory data, and agent post-training.
Researched medical-device and medical-AI companies; authored analysis reports.
Industry research covering medical devices and medical AI, including industry analysis reports and financial-model balancing.
Worked on GDE Mate, an AI agent that analyzes code repositories with RAG — automatically identifying file functions and dependencies and generating modification suggestions with sequence diagrams. Owned the interface design and implementation for refreshing sequence diagrams after human refinement, and designed a URL-based repository-query feature.
Multimodal temporal-signal processing and unobtrusive vital-sign monitoring.
Selected Projects
Borrowing ideas from NeRF, implicitly represents 3D electrophysiological signals with a Transformer and fuses semantic information with multi-view 2D modalities for 3D reconstruction. Uses a three-stage training pipeline (pre-training, calibration, fine-tuning); against baselines, achieves ≈6 dB higher PSNR in real-world settings with only half the parameters.
A UNet-based conditional GAN that segments periodic signals into individual cycles to reduce memory cost, lifts signals into a high-dimensional space to generate multi-view representations, and reconstructs signals end-to-end through single-cycle aggregation. Published in BSPC 2024.
Processed ECG signals (~1–50 µV) captured by the lab's self-developed armband on a single arm: analyzed noise characteristics for filtering and applied generative models for multi-view probabilistic reconstruction.
Biofeedback breathing-training system based on resonant-frequency extraction from single-source physiological signals (university innovation program, rated Excellent) · Edge AI platform for abnormal-signal detection with lightweight networks, F1 = 0.9 (National BME Innovation Design Competition, 2nd Prize) · Emotion decoding from high spatial-resolution surface electrophysiology (national innovation program).
Education
Honors & Awards
- 2026First-Class Scholarship for Outstanding Students, Sun Yat-sen University
- 20252nd Prize, 10th National BME Innovation Design Competition (Graduate Track)
- 2024First-Class Scholarship for Outstanding Students, Sun Yat-sen University
- 2023Outstanding Undergraduate Thesis, School of BME, Sun Yat-sen University
- 2023Exam-exempt Recommendation for Graduate Admission, SYSU BME
- 2023Second-Class Scholarship for Outstanding Students, Sun Yat-sen University
- 20212nd Prize, 6th National BME Innovation Design Competition (Undergraduate Track)
- 2020Academic Innovation Scholarship, Sun Yat-sen University
- 20191st Prize, SYSU Software Innovation & Entrepreneurship Competition