Library
-
Using LLMs to Bridge the Gaps in QA Test Plans at Firefox
Jan. 16, 2026Suhaib Mujahid, Marco Castelluccio, John Pangas, Ahmad AbdellatifGenerative AI / Open sourceThe study explores using LLMs to automatically generate test plans for Firefox features, aiming to reduce the manual effort and blind spots in traditional QA planning.
-
Automated Generation of Issue-Reproducing Tests by Combining LLMs and Search-Based Testing
Jan. 16, 2026Alberto Bacchelli, Marco Castelluccio, Konstantinos KitsiosGenerative AI / Open sourceThe paper introduces BLAST, a tool that combines large language models with search-based software testing to automatically generate issue-reproducing tests from issue-patch pairs, addressing the common absence of such tests in practice.
-
A Comparison of Conversational Models and Humans in Answering Technical Questions: the Firefox Case
Jan. 16, 2026Marco Castelluccio, Daniel Coutinho, Anita Sarma, Marco Gerosa, Caio Barbosa, Alessandro Garcia, Igor Steinmacher, Joao CorreiaGenerative AI / Open sourceThe study evaluates Retrieval-Augmented Generation (RAG) with LLMs for assisting Mozilla Firefox developers by comparing human responses, standard LLM, and RAG-enhanced LLM on real developer queries.
-
A Dataset of Performance Measurements and Alerts from Mozilla (Data Artifact)
Jan. 16, 2026Diego Elias Costa, Suhaib Mujahid, Marco Castelluccio, Gregory Mierzwinski, Mohamed Bilel BesbesOpen sourceThe paper introduces a publicly available dataset from Mozilla Firefox that addresses the lack of real-world data for studying performance regressions, combining performance measurements, expert-validated alerts, and rich metadata.
-
Impact of LLM-based review comment generation in practice: A mixed open-/closed-source user study
Jan. 16, 2026Suhaib Mujahid, Marco Castelluccio, Doriane Olewicki, Benjamin Mah, Leuson Da Silva, Arezou Amini, Sarra Habchi, Bram Adams, Foutse KhomhGenerative AI / Open sourceThe study evaluates RevMate, an LLM-based code review assistant, through a large-scale live user study at Mozilla and Ubisoft, analyzing over 587 patch reviews.
-
Towards Best Practices for Open Datasets for LLM Training
Jan. 13, 2025Stefan Baack, Stella Biderman, Aviya Skowron, Kasia OdrozekOpenness and AI / AI fairness, accountability, and transparencyBuilding on community insights from 30 AI dataset experts, this research paper distills best practices for creating open datasets for LLM training. The paper is a collaboration between Mozilla and EleutherAI.
-
From Skin to Screen Bodily Intgrity in the Digital Age
Nov. 20, 2024Júlia KeserűResearch examines the collection of “body-centric data” which has experienced a dramatic surge since the COVID-19 pandemic and the rise of sophisticated AI tools.
-
Incomplete Chronicles: Unveiling Data Bias in Maternal Health
Nov. 18, 2024Min’enhle NcubeAn ethnographic research into the datasets powering maternal healthcare app “DawaMom” used across Zambia and other Southern African countries
-
Public AI
Sept. 30, 2024Mark Surman, Nik Marda, Jasmine SunResponsible technologyA vision for a robust ecosystem of initiatives that promote public goods, public orientation, and public use throughout every step of AI development and deployment.
-
Unveiling the Potential of a Conversational Agent in Developer Support: Insights from Mozilla’s PDF.js Project
July 10, 2024Marco CastelluccioOpen sourceThis paper presents an investigation into whether AI can be leveraged to assist developers and guide open source community members. It introduces DevMentorAI, an LLM-based tool that uses a RAG approach to answer developer questions, which is then evaluated with a case study on Mozilla's PDF.js proje
