This talk presents a practical case study of how Free and Open Source Software (FOSS) is being used to support Open Science initiatives within the Indian research ecosystem. Through the creation of scalable, interactive computing environments using JupyterHub, Kubernetes, and open-source tools like Mercury, this work enables domain-specific scientific research in areas such as planetary science and meteorology. With a strong emphasis on reusability, accessibility, and automation, the project addresses common roadblocks faced by scientists in adopting FOSS-based workflows for research and collaboration.
Description:
Scientific research in India is increasingly adopting open standards, reproducibility, and collaboration as essential practices. However, challenges remain in onboarding researchers to Free and Open Source Software (FOSS) workflows that are often perceived as complex and fragmented. In this talk, I will share my experience designing and deploying a cloud-native, FOSS-powered platform to help scientists from diverse backgrounds engage in reproducible research.
Built using JupyterHub and Kubernetes, the platform offers scalable, secure, and domain-customizable computing environments tailored for scientific disciplines such as planetary science and meteorology. Customized Docker containers preloaded with scientific libraries allow researchers to spin up user-ready workspaces with minimal technical overhead.
To address the common barrier of converting computational work into shareable tools, the platform integrates Jupyter Server Proxy and the open-source library Mercury. This allows scientists to convert their notebooks into web applications directly within JupyterLab — turning complex research workflows into interactive apps for public sharing or internal collaboration. Mercury was further customized to fit within JupyterHub’s multi-user setup, and extended with RESTful APIs and JupyterLab extensions. This lets users intuitively manage their web apps (create, update, delete) through a graphical interface, without writing additional code.
This talk also reflects on the meta-challenges of driving FOSS adoption in scientific settings, including training researchers unfamiliar with software development, addressing infrastructure constraints, and designing systems that reduce friction while maintaining flexibility. The solution developed is already being piloted in academic and research contexts in India to support open, transparent science.
Attendees will gain insights into building and maintaining such a system, including:
Deploying JupyterHub on Kubernetes for scientific use cases
Creating Dockerized, domain-specific user environments
Seamlessly integrating JupyterLab with Mercury for science communication
Best practices in making FOSS more usable and accessible for scientists