Jiawei Zhang

Jiawei Zhang

Change is a good thing.

Welcome!

I am currently a research assistant at Westlake University, Prof. Yajie Wang’s Lab. And the main work direction is AI4Science and Enzyme mining.

I graduated from the School of Economics and Management of Civil Aviation University of China in 2020.

From 2022-2024, I worked with Dr. Hui Yao at Fresh Wind Biotechnologies, Inc. and was mainly responsible for algorithm development for cellular immunotherapy.

My research interest includes natural language processing and AI4Science (especially proteins and cancer).

You can find my CV here: Jiawei Zhang’s Curriculum Vitae

📝 Publications

TABR-BERT: an Accurate and Robust BERT-based Transfer Learning Model for TCR-pMHC Interaction Prediction

Jiawei Zhang, Wang Ma, Hui Yao

NeoMUST: an Accurate and Efficient Multi-Task Learning Model for Neoantigen

Wang Ma, Jiawei Zhang, Hui Yao

Lost in Tokenization: Context as the Key to Unlocking Biomolecular Understanding in Scientific LLMs

Kai Zhuang§, Jiawei Zhang§, et al.

  • 🔗Github repository arXiv 2025

  • In submission to International Conference on Learning Representations

PRIME: A Multi-Agent Environment for Orchestrating Dynamic Computational Workflows in Protein Engineering

Yuyang Zhou§, Jin Su§, Jiawei Zhang, et al. bioRxiv 2025

🖥️ Project

Multi-tool intergrated enzyme mining pipeline

Developer and researcher, supervised by Prof. Yajie Wang

  • Developed a multi-tool integrated workflow based on deep learning and vector search for enzyme mining in large-scale databases.

Energy-based TCR-pMHC binding prediction

Developer and researcher, supervised by Prof. Wengong Jin

  • Evaluating the predictive ability of metrics (pAE, pTM, etc.) generated by models based on multimer structure prediction for TCR-pMHC binding after energy function optimization.

ProteinChat: LLM-based protein function annotation

Developer and researcher, supervised by Prof. Fajie Yuan

  • Develop a chatbot that can map from protein sequence and structure to function through test-time scaling based on motifs, GO terms, and other features.

ProtAgent: An LLM-based biological agent

Developer and researcher, supervised by Prof. Fajie Yuan

  • Design and develop an LLM-based agent system that automatically executes corresponding AI models and bioinformatic tools based on the user’s needs.

SaprotHub: Making Protein Modeling Accessible to All Biologists 🔗Github repository

Assistant Developer, supervised by Prof. Fajie Yuan

  • Developed ColabSaprot and SaprotHub to support scientific research, allowing biologists to easily train and use Protein Language Models. SaprotHub is widely used for protein-related tasks, with wet lab experiments validating its results.

An LLM-based web3 agent

Researcher, supervised by Prof. Youzhi Zhang.

  • Design an agent that can decompose user requirements and then select and execute the appropriate tool for each step of the task.

TABR-BERT: TCR-pMHC Binding prediction base on transfer learning 🔗Github repository

Developer and researcher, supervised by Dr. Hui Yao

  • Predict the binding of TCR peptide-MHC-1 complexes by BERT-based deep learning model with transfer learning.

NeoMUST: Neoantigens identification base on multi-task learning 🔗Github repository

Developer and researcher, supervised by Dr. Hui Yao

  • Accurate identification of neoantigens by multi-task learning architecture with LSTM as the feature extractor.

🎖 Honors and Awards

  • 2017-2018 Renmin Third-Class Scholarship

💻 Internships