MSCS @ Northeastern University
Former Researcher @ Brigham and Women's Hospital / Harvard Medical School

I build machine learning systems, data infrastructure, and research pipelines for complex real-world problems.

My current interests focus on:

• AI systems and evaluation
• machine learning robustness under domain shift
• multimodal and imperfect-data learning
• scalable backend and data infrastructure
• human-in-the-loop AI systems
• reproducible ML workflows

Biomedical and neuroimaging applications are currently my primary problem domains, and my long-term focus is AI/ML systems and research engineering.

Current interests include:

• domain shift
• imperfect and heterogeneous data
• annotation-efficient learning
• self-supervised representation learning
• robustness across scanners/sites/distributions

Particularly interested in:

• representation learning
• sample selection
• generalization under distribution shift
• scalable biomedical ML pipelines

Working on:

• multi-agent AI systems
• evaluation pipelines
• semantic normalization workflows
• backend services and access-control systems
• scalable research tooling

Interested in how AI systems behave in noisy, real-world environments rather than benchmark-only settings.

• representation learning
• multimodal ML
• self-supervised learning
• medical imaging
• segmentation
• model evaluation
• robustness and generalization

• backend APIs
• authentication systems
• RBAC
• databases
• reproducible pipelines
• data versioning
• scalable tooling

• neuroimaging workflows
• MRI pipelines
• BIDS infrastructure
• large-scale biomedical datasets
• scientific Python ecosystem

Python
PyTorch
Flask
REST APIs
MongoDB
JWT / RBAC
Docker
Git/GitHub
Machine Learning
Deep Learning
Medical Imaging
Data Pipelines
Research Infrastructure

• algorithms and data structures
• ML systems engineering
• robust machine learning
• scalable backend systems
• software engineering practices
• deep learning for real-world deployment

• ML systems
• robust AI pipelines
• multimodal learning
• representation learning
• data-centric AI
• medical imaging AI
• AI infrastructure
• backend engineering for ML systems

I am particularly interested in research problems involving:

• distribution shift
• limited-label learning
• imperfect real-world datasets
• trustworthy AI systems
• evaluation and benchmarking
• scalable research tooling
• human-centered AI systems

Long-term goal: build AI systems that remain reliable under noisy, heterogeneous, and real-world conditions.

I prefer building systems that are:

• reproducible
• scalable
• inspectable
• modular
• deployable
• robust to imperfect data

I care less about leaderboard optimization and more about whether systems continue to work under realistic constraints.

• machine learning under domain shift
• medical imaging pipelines
• AI evaluation systems
• backend engineering for ML workflows
• transitioning from neuroscience/medicine into AI systems engineering
• reproducible research infrastructure

This GitHub is gradually being organized around:

1. ML systems
2. research engineering
3. data infrastructure
4. robust machine learning
5. biomedical AI pipelines
6. reproducible scientific workflows

Public repositories will mainly focus on:

• tooling
• pipelines
• infrastructure
• methods
• reproducible engineering workflows

• LinkedIn: https://www.linkedin.com/in/mengyuan-ding
• GitHub: https://github.com/Jessy-Ding
• Email: ding.mengyu@northeastern.edu