POSE: Phase I: ModularML: An Open-Source Ecosystem for Modular and Extensible Machine Learning

This Pathways to Enable Open-Source Ecosystems (POSE) project will expand access to machine learning (ML) for scientists and engineers by advancing ModularML, an open-source, research-oriented machine learning framework, into a sustainable, community-driven ecosystem. While machine learning is transforming the pace and scope of scientific discovery, many existing tools require computer science or software engineering backgrounds and are difficult to adapt to different research areas. This project addresses these gaps by strengthening ModularML’s usability, extensibility, and adoption. The new open-source ecosystem (OSE) will broaden access to machine learning tools for a wide range of domain scientists. It will help accelerate innovation in areas critical to national priorities, including energy, infrastructure, and advanced manufacturing. By providing a reproducible, transparent, and user-friendly machine learning framework, ModularML will enhance educational and training outcomes, support workforce development in computational science and engineering, and promote open science practices. Through partnerships with academic and industrial collaborators, ModularML will serve as a template for how research-focused machine learning tools can evolve into community-driven ecosystems that increase access to innovation. This POSE project will lay the foundation for transitioning ModularML into a sustainable and community-driven OSE. ModularML is a backend-agnostic framework that enables construction and execution of machine learning workflows through a graph-based architecture with modular components for data processing, feature engineering, model definition, and multi-stage training. The project team will scope ModularML’s readiness for the OSE transition by: (1) evaluating its position in the machine learning software landscape and tracking community engagement; (2) developing governance and contribution policies for sustainable growth; (3) engaging and training domain researchers and contributors through workshops, forums, and onboarding resources; and (4) formalizing a plugin system and development roadmap to encourage community-driven, domain-specific extensions. Collectively, these efforts will establish the technical and organizational foundations for a scalable OSE that enhances accessibility of machine learning tools and contributes to broader best practices in software modularity, governance, and open-source sustainability. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2550230 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Chao Hu | Institution: University of Connecticut, STORRS, CT | Award Amount: $299,995 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2550230 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2550230.html