Thu NguyenPh.D.

Computational scientist · AI for biology

Designing proteins with deep learning, biological insight, and computational precision.

I’m Thu Nguyen, a computational scientist developing deep-learning & machine-learning methods for protein design, deimmunization, structure prediction, and drug discovery, from research prototypes to scalable GPU and cloud workflows.

Explore my workView research
Computer Science
Ph.D.Computer Science
NetMHC score reduction
70%+NetMHC score reduction
sequence recovery
60%+sequence recovery
listed publications
5listed publications
Highlight
Industry + National-lab Research

Selected results reported in my resume performance depends on the model, protein, and evaluation setting.

From structure prediction to therapeutic design

My work connects computer science with the physical and biological constraints that shape proteins. I build and adapt deep-learning models, design optimization strategies, and create scalable research workflows that help teams explore better protein sequences, binders, and structural hypotheses.

Selected impact

Research that changes how sequences get designed

Industry · Protein engineering

Optimizing protein sequences for lower immunogenicity without losing function

Implemented MCMC optimization combining NetMHC predictions with inverse-folding scores from IF1 and ProteinMPNN to explore sequence designs under competing biological constraints.

70%+

reduction in NetMHC score on monomer Griffithsin

10%

increase in reported TM/activity metric

Method details
MCMCNetMHCESM-IF1ProteinMPNNMulti-objective optimization

Described only at the level of detail present in Thu's public résumé.

Industry · Transformers

Learning fitness-aware protein sequences from structure, templates, and MSAs

Designed and trained a transformer-based approach that integrates templates and multiple-sequence alignments for sequence design and fitness prediction.

60%+

sequence recovery on validation data

~45-50%

comparison range reported for IF1 and ProteinMPNN

0.7

reported correlation with Ides activity, vs. 0.5 for the referenced ESM-IF1 score

Method details
TransformersMSAFitness predictionPyTorchEvaluation design

Ph.D. research · Structural biology

Advancing protein structure prediction and design with geometric deep learning

Reproduced and extended AlphaFold-style monomer and multimer training, improved SE(3)-Transformer attention, and developed deep-learning methods for peptide and binding-oriented design across CUDA and ROCm environments.

Up to 30%

precision improvement reported on an n-body dataset after adding self-attention kernels

10%

improvement in docking accuracy reported after implementing Generalized Born scoring for FTMap

Method details
AlphaFoldSE(3)-TransformerGeometric learningCUDAROCmDistributed training
Read the research story

A scientist who builds

  1. 01

    Frame

    Translate the biological objective into measurable computational constraints.

  2. 02

    Model

    Select, reproduce, adapt, or design the learning and optimization approach.

  3. 03

    Scale

    Build repeatable GPU and cloud training/inference workflows.

  4. 04

    Validate

    Compare against strong baselines and interpret results in biological context.

Insmed Inc. · 2023–Present

Computational protein design in industry

At Insmed, I develop computational approaches for protein design and deimmunization. My work spans sequence optimization, inverse folding, transformer training, fitness prediction, pH-sensitive binder design, and cloud-based scientific computing.

Career

Research timeline

  1. Dec 2023 – Present

    Insmed Inc.

    Scientist

    Computational protein design and deimmunization: sequence optimization, inverse folding, transformer-based fitness prediction, pH-sensitive binder design, and cloud-based scientific computing.

  2. Aug 2020 – Dec 2023

    Stony Brook University, Applied Bio Computation Lab

    Research Assistant

    Doctoral research in geometric deep learning and structure prediction, including AlphaFold reproduction and SE(3)-Transformer extensions.

  3. Jun 2019 – Jun 2021

    Brookhaven National Laboratory

    Research Assistant

    Crystallography data processing: diffraction-image classification, lossy compression, and structure-factor analysis.

  4. Sep 2018 – Dec 2018

    Stony Brook University, Computational Bio Lab

    Research Assistant

    Sequence alignment performance and parallelization.

  5. Feb 2016 – Aug 2018

    Vietnam National University, University of Science

    Research Assistant

    Early research in applied machine learning, NLP, and statistical forecasting.

Publications

Selected publications

View all publications

A simple technique to classify diffraction data from dynamic proteins according to individual polymorphs

2022
Nguyen, T., Phan, K. L., Kozakov, D.+ 7 more

Nguyen, T., Phan, K. L., Kozakov, D., Gabelli, S. B., Kreitler, D. F., Andrews, L. C., Jakoncic, J., Sweet, R. M., Soares, A. S., Bernstein, H. J.

Acta Crystallographica Section D: Structural Biology, 78(3), 268-277

View DOI (opens in a new tab)
CrystallographyStructural Biology

Assessing the binding properties of CASP14 targets and models

2021
Egbert, M., Ghani, U., Ashizawa, R., Nguyen, T.+ 7 more

Egbert, M., Ghani, U., Ashizawa, R., Kotelnikov, S., Nguyen, T., Desta, I., Hashemi, N., Padhorny, D., Kozakov, D., Vajda, S.

Proteins: Structure, Function, and Bioinformatics, 89(12), 1922-1939

View DOI (opens in a new tab)
Structural BiologyProtein-Ligand BindingDocking

Conservation of binding properties in protein models

2021
Egbert, M., Porter, K. A., Ghani, U., Nguyen, T.+ 5 more

Egbert, M., Porter, K. A., Ghani, U., Kotelnikov, S., Nguyen, T., Ashizawa, R., Kozakov, D., Vajda, S.

Computational and Structural Biotechnology Journal, 19, 2549-2566

View DOI (opens in a new tab)
Structural BiologyProtein-Ligand Binding

Entrepreneurship

Applying AI beyond the lab

As a founder of Future Readers, I help explore how AI can create personalized learning and storytelling experiences for children. The work reflects my interest in translating emerging technology into thoughtful products for real people.

Visit Future Readers

Let’s turn difficult scientific questions into systems we can test.

I’m interested in senior opportunities and collaborations across computational biology, protein design, scientific machine learning, and AI-enabled drug discovery.

Start a conversationConnect on LinkedIn