I'm an undergraduate researcher at Jiangnan University, School of AI & Computer Science (Class of 2027). My work spans Multimodal Learning, AI Safety, LLM Reasoning, and AI for Science (Computational Biology).
- Published / submitted 4 papers as first or co-first author (1 CCF-B, 3 CCF-C)
- iGEM 2025 Gold Medalist — presented in Paris as dry-lab lead
Multimodal Learning & VLMs · Generative AI Safety & Deepfake Detection
LLM Reasoning & Evaluation · AI for Science (Protein Design / Genomics)
| # | Title | Venue | Role | Status |
|---|---|---|---|---|
| 1 | MambaGuard: A CLIP-Mamba Approach for OOD Generated Image Detection | PRCV 2026 (CCF-C) | Co-first Author | Accepted |
| 2 | SHINE: A Neuro-Symbolic Approach to Language-Driven Travel Planning | ICAPS 2026 (CCF-B) | First Author | Under Review |
| 3 | MLLM-based Image Forgery Detection and Localization | ICIC 2026 (CCF-C) | First Author | Under Review |
| 4 | Response-Pattern Enhanced Item Response Theory for Large Language Model Evaluation | IJCNN 2026 (CCF-C) | First Author | Under Review |
|
iGEM 2025 Gold Medal — Antimicrobial Peptide de novo Design Platform Transformer-VAE + Latent Diffusion architecture fusing protein language models (ESM-2 / ProtT5 / Ankh) for conditional AMP generation with 6 generation modes.
|
Universal Biomolecular Rational Design System End-to-end protein-ligand complex design covering small molecules, peptides, and DNA/RNA aptamers. Optimized pLDDT from 0.55 to 0.75+.
|
|
Genomic Language Model Domain Adaptation LoRA fine-tuning of Evo2-1B on BioNeMo Framework with Megatron-LM distributed preprocessing. Full bio-sequence deep learning pipeline.
|
Neuro-Symbolic Travel Planning Hybrid LLM + symbolic reasoning architecture with constraint-aware DFS. Achieved 77.27% pass rate on ChinaTravel benchmark (SOTA).
|
Deep Learning & LLMs
Architectures & Techniques
AI for Science
Languages & Tools
