Pfizer logoCurrently an Extern @Pfizer
dayoon Jung portrait

Dayoon Jung: MPhys Physics Graduate

I'm a theoretical physics graduate specializing in statistical deep learning and machine learning engineering for MRI. I create impactful AI solutions.

View my workGet in touch

Works samples

I develop AI platforms and classification pipelines that bridge physics and machine learning, focusing on MRI and healthcare applications.

  • NeuroCVR-AI: Synthetic MRI Benchmarking and ML Engineering Platform
    NeuroCVR-AI: Synthetic MRI Benchmarking and ML Engineering Platform

    ·

    NeuroCVR-AI: Synthetic MRI Benchmarking and ML Engineering Platform

    Built an independent end-to-end ML/R&D platform for synthetic MRI response benchmarking. Simulated physiological stimulus-driven MRI time-series data for controlled model evaluation. Implemented voxelwise statistical estimation of response magnitude and timing maps. Generated NIfTI outputs, quantita

    PythonDockerMakefilepytestRuffFastAPIAWS EC2MLflowYAMLCLI
  • ML Classification Pipeline: Childhood Depression Risk, UNICEF MICS Malawi

    ·

    ML Classification Pipeline: Childhood Depression Risk, UNICEF MICS Malawi

    Built a scikit-learn classification pipeline for 87-feature mixed-type survey data. Used preprocessing pipelines for imputation, encoding and feature scaling. Optimised a Random Forest classifier using RandomizedSearchCV. Interpreted model behaviour using feature importances and logistic regression

    scikit-learnRandom ForestRandomizedSearchCV

About me

I'm a theoretical physics graduate specializing in statistical deep learning and machine learning engineering for MRI. I create impactful AI solutions.

MPhys Theoretical Physics graduate with experience in statistical deep learning, physics-informed neural networks and ML engineering for quantitative MRI. Skilled in Python/PyTorch, simulation-based validation and reproducible model evaluation.

Externships

Pfizer Advanced: AI-Powered Document Insights & Data Extraction Externship

Pfizer

Experience

First-Author Poster Presenter, SINAPSE ASM Edinburgh 2026

SINAPSE ASM Edinburgh · Jun 2026

Project Reporter

EconoVogue Societe, Remote · Sept 2023 – May 2024

Processing Lab Chemist

Engineering for Change, Edinburgh · Oct 2023 – Apr 2024

Education

University of Edinburgh

MPhys Theoretical Physics (2:1 Honours) · Class of 2026

BVIS Hanoi

A-levels (Mathematics A*, Chemistry A*, Physics A) · Class of 2022

Skills

PythonPyTorchsimulation-based validationreproducible model evaluationstatistical deep learningphysics-informed neural networksML engineeringscikit-learnimputationencodingfeature scalingRandom ForestRandomizedSearchCVfeature importanceslogistic regressionSQLGit/GitHubLinuxDockerFastAPITyperMLflowAWS EC2pytestRuffMakefileAPI testingNumPySciPyPandasMatplotlibBland–Altman analysisPearson correlationRMSEMAESSIMbias analysisGaussian KDEMRI time-series analysisfunctional MRIstructural MRINIfTI/BIDSFSLFreeSurfer/FastSurfervoxelwise modellingimage registrationtissue segmentationanatomical maskingautomatic differentiationODE modellingsimulation designparameter recoveryhyperparameter searchmodel comparisonSIREN networks
Back to works

NeuroCVR-AI: Synthetic MRI Benchmarking and ML Engineering Platform

Built an independent end-to-end ML/R&D platform for synthetic MRI response benchmarking. Simulated physiological stimulus-driven MRI time-series data for controlled model evaluation. Implemented voxelwise statistical estimation of response magnitude and timing maps. Generated NIfTI outputs, quantita

PythonDockerMakefilepytestRuffFastAPIAWS EC2MLflowYAMLCLI
NeuroCVR-AI: Synthetic MRI Benchmarking and ML Engineering Platform

Overview

Built an independent end-to-end ML/R&D platform for synthetic MRI response benchmarking. Simulated physiological stimulus-driven MRI time-series data for controlled model evaluation. Implemented voxelwise statistical estimation of response magnitude and timing maps. Generated NIfTI outputs, quantita

View all works
Back to works

ML Classification Pipeline: Childhood Depression Risk, UNICEF MICS Malawi

Built a scikit-learn classification pipeline for 87-feature mixed-type survey data. Used preprocessing pipelines for imputation, encoding and feature scaling. Optimised a Random Forest classifier using RandomizedSearchCV. Interpreted model behaviour using feature importances and logistic regression

scikit-learnRandom ForestRandomizedSearchCV

Overview

Built a scikit-learn classification pipeline for 87-feature mixed-type survey data. Used preprocessing pipelines for imputation, encoding and feature scaling. Optimised a Random Forest classifier using RandomizedSearchCV. Interpreted model behaviour using feature importances and logistic regression

View all works