Lowering Barriers in Open Source: AI-Supported Onboarding with CommUnityBuddy
MCLD 2002 | Sun 09 Aug 3:25 p.m.–3:45 p.m.
Presented by
-
Pedro Oliveira
https://www.linkedin.com/in/pedro-oliveira-3698141a3/
Ph.D. student in Informatics with a focus on open-source software ecosystems, governance, and AI-supported collaboration. My research examines how communities structure decision-making, define roles, and sustain participation over time. I investigate governance models, contributor behavior, and project health through a combination of socio-technical theory, large-scale GitHub data analysis, and explainable machine learning. I develop predictive models to better understand developer turnover, contributor trajectories, and long-term sustainability in open-source projects. My broader goal is to bridge research and practice by designing tools, models, and frameworks that improve how software communities grow, collaborate, and sustain themselves over time.
-
I am the Associate Director for Graduate Programs in the School of Informatics, Computing, and Cyber Systems at Northern Arizona University. I earned my Ph.D. in Computer Science from the University of São Paulo in 2015 and was a visiting scholar at the University of California, Irvine (2013-2014).
My research bridges Software Engineering and Computer-Supported Cooperative Work, with a focus on human aspects of software development. I study how to better support newcomers in Open Source Software communities and how emerging technologies—especially foundational models—are reshaping education and software development. More broadly, my work spans empirical software engineering, OSS sustainability, and mining software repositories.
Pedro Oliveira
https://www.linkedin.com/in/pedro-oliveira-3698141a3/
Ph.D. student in Informatics with a focus on open-source software ecosystems, governance, and AI-supported collaboration. My research examines how communities structure decision-making, define roles, and sustain participation over time. I investigate governance models, contributor behavior, and project health through a combination of socio-technical theory, large-scale GitHub data analysis, and explainable machine learning. I develop predictive models to better understand developer turnover, contributor trajectories, and long-term sustainability in open-source projects. My broader goal is to bridge research and practice by designing tools, models, and frameworks that improve how software communities grow, collaborate, and sustain themselves over time.
I am the Associate Director for Graduate Programs in the School of Informatics, Computing, and Cyber Systems at Northern Arizona University. I earned my Ph.D. in Computer Science from the University of São Paulo in 2015 and was a visiting scholar at the University of California, Irvine (2013-2014).
My research bridges Software Engineering and Computer-Supported Cooperative Work, with a focus on human aspects of software development. I study how to better support newcomers in Open Source Software communities and how emerging technologies—especially foundational models—are reshaping education and software development. More broadly, my work spans empirical software engineering, OSS sustainability, and mining software repositories.
Abstract
Open source communities depend on new contributors, but onboarding remains one of the biggest barriers to participation. Newcomers often struggle to find where to start, understand project codebases, navigate documentation, and interact with the community. These challenges frequently lead to frustration and early drop-off. This talk presents CommUnityBuddy, an AI-powered assistant to support newcomer onboarding in open source scientific software projects. The system uses large language models with project specific data, such as documentation, issues, pull requests, and chat logs, to provide contextual real-time guidance.
The goal is not to replace human mentorship, but to scale it. By answering repetitive questions, and guiding contributors through common workflows, the assistant helps reduce maintainer workload while giving newcomers faster and more accessible support. The talk will cover key onboarding challenges, how AI can be used to support newcomers, the design of a retrieval-augmented system, and early evaluation insights from studies.
This work shows how AI can improve newcomers onboarding without disrupting community practices, offering a path toward more inclusive and sustainable open source participation.
Open source communities depend on new contributors, but onboarding remains one of the biggest barriers to participation. Newcomers often struggle to find where to start, understand project codebases, navigate documentation, and interact with the community. These challenges frequently lead to frustration and early drop-off. This talk presents CommUnityBuddy, an AI-powered assistant to support newcomer onboarding in open source scientific software projects. The system uses large language models with project specific data, such as documentation, issues, pull requests, and chat logs, to provide contextual real-time guidance.
The goal is not to replace human mentorship, but to scale it. By answering repetitive questions, and guiding contributors through common workflows, the assistant helps reduce maintainer workload while giving newcomers faster and more accessible support. The talk will cover key onboarding challenges, how AI can be used to support newcomers, the design of a retrieval-augmented system, and early evaluation insights from studies.
This work shows how AI can improve newcomers onboarding without disrupting community practices, offering a path toward more inclusive and sustainable open source participation.