Senior Algorithm Engineer working on reliable visual recognition systems, with a focus on OOD Detection, Failure Prediction, and production deployment.
Portfolio / SUREPlus / SSB-OSR
I develop classification, face recognition, Re-ID, detection, and segmentation systems from training and evaluation through optimization and deployment. My research asks a practical question: how can a model recognize when its own prediction is unreliable?
Reliable classification OOD detection Failure prediction
Open-set recognition Metric learning Production vision
| Work | Contribution | Status |
|---|---|---|
| From Misclassifications to Outliers | Joint reliability assessment for classification | TPAMI, under review, first author |
| Zigzag-ViT | Prompted zigzag interaction for reliable Vision Transformers | Pattern Recognition, under review, first author |
| EdgeCrafter | Lightweight high-performance models for edge devices | TMLR, accepted, co-first author |
| Automated 3D Building Reconstruction from a Single Satellite Image | Automated reconstruction for mapping workflows | Published, first author |
- Senior Algorithm Engineer at J&T Express, Jun 2026 - present
- Computer Vision Algorithm Engineer at Hefei Intellindust Information Technology Co., Ltd., Jul 2023 - Jun 2026
- B.E. and M.E. in Surveying and Mapping Engineering from Wuhan University
- Team leader and champion of the Open-set Recognition track, ECCV 2024 Workshop OOD-CV Challenge
- Six invention patent applications in substantive examination
For the full experience, research notes, and project links, visit my personal homepage.