Jieke(Jack) Wu

Jieke(Jack) Wu

Graduate Student at King Abdullah University of Science and Technology (KAUST)

πŸ‘¨πŸΌβ€πŸŽ“ About me

I’m Jieke (Jack) Wu (武杰克), a graduate student at King Abdullah University of Science and Technology (KAUST), pursuing MS/PhD in Computer Science. I am a member of the Structural and Functional Bioinformatics (SFB) Research Group, led by Prof. Xin Gao, where I focus on cutting-edge research at the intersection of computational biology, artificial intelligence, and protein design.

My research journey began at the Department of Life Sciences and Medicine, University of Science and Technology of China, where I earned my B.S. in Biological Technology with a strong academic foundation. This interdisciplinary background has uniquely positioned me to bridge the gap between traditional biological research and modern computational approaches.

My current research interests focus on:

  • Protein Design: Developing advanced generative models for de novo protein design and structure prediction
  • Multi-Agent Systems: Exploring intelligent agent frameworks for complex biological and medical tasks
  • AI + Drug Discovery: Leveraging artificial intelligence to accelerate pharmaceutical development and drug design
  • AI + Healthcare: Creating AI solutions for medical diagnosis, treatment optimization, and personalized medicine

I am actively researching generative models including diffusion models and flow matching techniques, as well as developing intelligent agents for various applications in computational biology and healthcare.

If you are interested in collaborating on research projects or would like to discuss potential opportunities, please feel free to contact me at jieke.wu@kaust.edu.sa.

πŸ“– Educations

πŸ“‘ Publications

  • Jieke Wu, Wei Huang, Mingyuan Bai, Xiaoling Hu, Yi Duan, Wuyang Chen. β€œTraining-free Design of Augmentations with Data-centric Principles.” ICML 2024 Workshop AI4Science. This work introduces a novel framework for evaluating data augmentation strategies without requiring expensive model training, significantly reducing computational costs while maintaining accuracy.

  • Tinghui Wu$^\dag$, Jieke Wu$^\dag$, Zijun Zhang, Wuyang Chen. β€œTraining-free Design of Deep Networks as Ensembles of Clinical Experts.” Preprint on medRxiv. This collaborative work presents TEACUP, a training-free evaluation framework that enables the creation of AI model ensembles for clinical applications, improving prediction robustness and uncertainty quantification.

πŸ§ͺ Research Experience

🧬 Evaluation Task Design for GenomeOcean

Northwestern University, Prof. Han Liu
Remote (US)
Research Assistant (09/2024 – 11/2024)

This project focused on developing comprehensive evaluation metrics for assessing the quality and biological plausibility of DNA sequences generated by the GenomeOcean model. My contributions included:

  • Designed novel evaluation metrics to assess the novelty and biological plausibility of generated DNA sequences, ensuring they meet real-world biological constraints
  • Conducted comparative analysis of open reading frame (ORF) length distributions between model-generated sequences and real DNA sequences, providing quantitative measures of sequence consistency
  • Analyzed codon bias patterns to evaluate the biological realism of generated sequences, comparing them against established genomic databases
  • Demonstrated superior performance of GenomeOcean compared to state-of-the-art models, achieving significant improvements in both ORF length distribution accuracy and codon bias consistency
  • Developed automated evaluation pipelines that can be applied to assess other DNA generation models, contributing to the broader field of computational genomics

🧠 Biology Learning Assistant Powered by GenAI

Northwestern University, Prof. Han Liu
Remote (US)
Research Assistant (07/2024 – 09/2024)

This innovative project aimed to revolutionize high school biology education through the application of large language models. My key contributions included:

  • Developed fine-tuned large language models using Low-Rank Adaptation (LoRA) techniques, specifically optimized for high school biology education content
  • Created comprehensive training datasets by processing biology textbooks using advanced OCR tools and integrating with Qwen2-API for content enhancement
  • Established a new benchmark using 20 years of National High School Biology Competition questions, providing a robust evaluation framework for educational AI systems
  • Demonstrated superior performance of the fine-tuned Qwen2 model compared to baseline student performance, showing significant improvements in accuracy and explanation quality
  • Implemented adaptive learning features that can personalize educational content based on student performance and learning patterns

πŸ’» Training-free Design of Deep Networks (TEACUP)

Natera, Dr. Zijun Zhang
Simon Fraser University, Prof. Wuyang Chen
Remote (US)
Research Assistant (08/2024 – 05/2025)

This groundbreaking project developed TEACUP (Training-free Evaluation of AI Clinical Understanding and Performance), a novel framework that revolutionizes how we evaluate clinical AI systems:

  • Developed TEACUP framework that provides training-free evaluation of clinical AI network performance, eliminating the need for expensive model training cycles
  • Achieved 90% reduction in computational costs while simultaneously improving clinical task performance across multiple medical imaging datasets
  • Implemented ensemble modeling strategies that simulate the collaborative decision-making of multiple human experts in clinical settings
  • Enhanced prediction robustness and uncertainty quantification, crucial for medical applications where reliability is paramount
  • Created scalable evaluation protocols that can be applied across different clinical domains and imaging modalities

🧬 Hierarchical Transformer for Genomics

Cedars-Sinai Medical Center, Prof. Zijun Zhang
Simon Fraser University, Prof. Wuyang Chen
Remote (US)
Research Assistant (03/2024 – 08/2024)

This project explored the application of advanced deep learning techniques to uncover hidden patterns in DNA sequences:

  • Investigated hidden patterns in DNA sequences using hierarchical transformer architectures, enabling both local and global sequence analysis
  • Improved model performance by developing methods that integrate global context with localized genomic information, achieving better prediction accuracy
  • Enhanced understanding of LLMs and gained proficiency with HuggingFace and other deep learning toolboxes for genomic applications
  • Developed novel attention mechanisms that can capture both short-range and long-range dependencies in DNA sequences
  • Applied transfer learning techniques to leverage pre-trained models for specific genomic prediction tasks

πŸ–₯ Training-free Data-centric Augmentations

UC Berkeley, Dr. Wuyang Chen
Remote (Canada)
Research Assistant (06/2023 – 02/2024)

This research focused on developing training-free methods for evaluating and designing data augmentation strategies:

  • Developed innovative metrics for data quality evaluation based on deep learning theory, providing quantitative measures of dataset characteristics
  • Introduced training-free data augmentation design principles that reduce computational costs while maintaining model performance
  • Improved medical image segmentation performance across multiple datasets, demonstrating the effectiveness of the proposed methods
  • Investigated the relationship between data covariance properties and image recognition accuracy, providing theoretical insights
  • Published findings in the ICML 2024 Workshop AI4Science, contributing to the broader AI4Science community

πŸ€ Isolation of Bacteriophages Targeting Gut Bacteria

University of Science and Technology of China, Prof. Yi Duan
Hefei, China
Research Assistant (01/2023 – 05/2024)

This project addressed critical challenges in gut microbiome research through innovative bacteriophage isolation techniques:

  • Isolated Akk-targeting phages from wastewater samples, creating a comprehensive phage library for gut microbiome manipulation
  • Developed novel purification protocols that maintain phage viability while removing contaminants
  • Characterized phage-host interactions to understand the specificity and efficiency of the isolated phages
  • Received outstanding rating as a school-level research initiative, recognizing the project’s innovation and potential impact

🦟 Biodegradable Needles for Transdermal Delivery

Suzhou Institute for Advanced Research, Prof. Xiaorong Xu
Suzhou, China
Research Assistant (11/2022 – 09/2023)

This interdisciplinary project combined materials science, mechanical engineering, and biomedical applications:

  • Simulated finite elements using COMSOL and Abaqus to optimize needle design for deep tissue penetration
  • Optimized long microneedles for treating deep tissue infections, considering both mechanical properties and biological compatibility
  • Developed novel injection molding techniques for economically producing complex microneedle structures
  • Conducted mechanical testing to ensure needles can penetrate skin layers while maintaining structural integrity
  • Received recognition as an outstanding school-level project for its innovative approach to drug delivery

🦠 Isolation of Cyanobacteria and Cyanophages from Lake Chaohu

Laboratory of Biochemistry & Structural Biology, Prof. Congzhao Zhou
Hefei, China
Research Assistant (09/2022 – 06/2023)

This environmental microbiology project contributed to our understanding of freshwater ecosystems:

  • Isolated three distinct strains of cyanobacteria and their corresponding cyanophages from Lake Chaohu water samples
  • Conducted comprehensive genomic analysis to determine taxonomic classification and ecological roles
  • Contributed to freshwater ecosystem understanding and potential applications in environmental monitoring

πŸ’» Internships

🧬 Protein Generation Model Research

MoleculeMind, Shanghai
Shanghai, China
Research Intern (11/2024 – 06/2025)

This cutting-edge internship focused on developing next-generation protein generation models using advanced AI techniques:

  • Developed unconditional protein generation models using diffusion techniques in SE(3) space, enabling de novo protein design with unprecedented accuracy
  • Investigated Flow Matching techniques in SE(3) space to enhance the generative capabilities of protein models, improving sampling efficiency and quality
  • Explored conditional protein generation tasks including De novo design and binder design, advancing the model’s applicability in drug discovery and synthetic biology
  • Contributed to the development of tools that can accelerate drug discovery processes and protein engineering applications

πŸŽ– Honors and Awards

  • Outstanding Graduates of the Class of 2025 from the University of Science and Technology of China (2025) - Recognized for academic excellence and research contributions
  • Outstanding School-Level Project: Undergraduate Innovation and Entrepreneurship Training Program (2024) - For innovative bacteriophage isolation techniques
  • Outstanding School-Level Project: Undergraduate Research Project (2023) - For biodegradable microneedle development
  • 8th National University Life Science Competition, A Prize (2023) - For cyanobacteria and cyanophage research
  • Outstanding Undergraduate Scholarship (2024, 2023, 2022, 2021) - Consistent academic excellence over four years

🧭 Skills

Programming Languages

  • Python: Advanced proficiency in scientific computing, machine learning, and bioinformatics applications
  • C/C++: Strong foundation in systems programming and performance-critical applications
  • Matlab: Experience in numerical computing and signal processing

Frameworks and Tools

  • HuggingFace: Expertise in transformer models, fine-tuning, and deployment of large language models
  • PyTorch Lightning: Advanced deep learning workflows and distributed training
  • PyTorch: Comprehensive experience in deep learning model development and optimization
  • Git: Version control and collaborative development practices
  • $\LaTeX$: Professional document preparation and academic writing

πŸŽ₯ Personal Interests

  • Anime: As a pastime in my spare time, I enjoy watching Japanese anime spanning various genres including romance, sports, mythology, and science fiction. This hobby helps me maintain creativity and cultural awareness while providing relaxation from intensive research work.

  • Interdisciplinary Learning: I am passionate about exploring the intersections between different fields, particularly how computational methods can advance biological research and medical applications.

  • Open Source Contribution: I believe in the power of collaborative development and actively contribute to open-source projects in the AI and bioinformatics communities.