Jethro Au 💽
Jethro Au

Healthcare Data Rese | AI/ML Research Engineer

Harvard University
Harvard Medical School / Massachusetts General Hospital AI & Robotics Lab

About Me

Hey there! I’m a machine learning engineer and data scientist with an interest in generative drug design, explainable medical language models, and multi-modal agents. I also enjoy researching interventions and prediction systems combating drug-resistant pathogens and therapeutic resistant cell mutations.

Some projects I’ve worked on include: generative de-novo peptide design for HIV using AlphaFold, recursive language models to reduce hallucinations, multi-modal generative agents, infection surveillance mechanisms using IoT sensors and digital twins, and more.

You may also find me mixing music sometimes, or I’d like to refer as - “manual” audio-signal processing.

Please reach out to collaborate 😃

Download CV
Interests
  • AI for Drug Discovery
  • Medical Language Models
  • Antimicrobial Resistance and Therapeutic Resistance
  • Multi-modal Agents
Education

  • MS in Healthcare Data Science

    Harvard University

  • TGP Fellow in Big Data Technology (MS)

    Hong Kong University of Science and Technology

  • BS Industrial Engineering

    Northwestern University

📚 My Research

AI/ML engineer specializing in generative drug design, medical language models, and data methods against drug and therapeutic resistant pathogens and cancer cells.

I also spend my time building ventures for healthcare companies such as building medical devices using cationic polymers effective against MDROs, drug-discovery with generative AI, and infection surveillance systems!

Please reach out to collaborate 😃

Featured
Cross-domain task generalization of graphs with language embeddings
Graph Neural Networks

Cross-domain task generalization of graphs with language embeddings

REPO LINK HERE Scaling and generalizing cross-domain tasks amongst graph datasets remains a challenge due to the variability in node features, edge-based relationships, and the inherit challenges for transfer-learning amongst graphs. The aim of this project is to explore the capabilities of using language embeddings to achieve task generalizations across different graph structures and build models that could learn cross-domain relationships. By evaluating the performance of trained graph neural networks across different language embeddings, the authors evaluate the effectiveness of various encoding architectures. Contrary to expectations, the simpler word2vec achieved greater performance compared to the E5-Small-V2 and GraphAlign pre-trained embeddings. Finally, the author discusses limitations and the conclusiveness of the study and discusses future research directions in unifying cross-domain graphs with scalable architecture.

Precision Design of Cyclic Peptides using AlphaFold
AlbumGen: An exploration into Multimodality with LLMs
Financial considerations in primary healthcare after COVID-19
AlphaFold2 for Cyclic Peptide Design - A Deep-dive into Generative Drug Design (2/2)
Recent Work
Jethro Au (2024). Precision Design of Cyclic Peptides using AlphaFold. Precision Design of Cyclic Peptides using AlphaFold.
PDF Cite Project Source Document
Jethro Au, Lai Chun Yu (HKUST) (2024). AlbumGen: An exploration into Multimodality with LLMs. AlbumGen - Image-to-Music Generation with Textual Intermediaries.
Cite Code Dataset Video
Jethro Au, Chloe Dian Yu (HKUST), Carol Wing Kwan Chow (Gleneagles Hospital, HK), Patrick Pak Ling Lam (Biomedical Impact Association, HK) (2024). Financial considerations in primary healthcare after COVID-19. Springer Nature.
Cite The Handbook of Primary Healthcare
Latest Blogs/Articles