SciPy India Community Call #1 – July 2025#
Agriya Khetarpal, Aditi Juneja, and Srihari Thyagarajan
This blog post is our account of our first SciPy India community call held on July 26th, 2025. Here, we describe the call as it happened, including updates from the community, the speakers and their presentations, and our discussions on future plans.
Details#
Date: July 26th, 2025
Time: 10:00 to 13:00 IST
Platform: Jitsi Meet (online)
Agenda#
We had a packed agenda for our first community call. Here’s a brief overview of what we covered:
Welcome and introductions
Talk 1: Automating ML with PyCaret by Manjunath Janardhan
Talk 2: marimo: an open-source, reactive Python notebook by Srihari Thyagarajan
Breakout sessions, discussions, and networking
Lightning talk 1: Introduction to Causal Machine Learning and
pgmpyby Mohammad RazakLightning talk 2: Subgraph Matching: The Needle-in-a-Network Problem by Varuni H K
Closing remarks and next steps
Highlights#
Manjunath Janardhan’s presentation on automating machine learning with PyCaret was insightful and demonstrated practical applications of the library, and how it makes machine learning accessible to a broader audience. Building up from machine learning from first principles, he demonstrated how PyCaret streamlines the typical machine learning pipeline — from data preprocessing and feature engineering to model training, evaluation, and deployment; abstracting away much of the complexity involved in building and deploying machine learning models, for non-experts and experts alike, whether for learning or for production. This was followed by Manjunath’s live demo of PyCaret’s capabilities using a sample dataset that provides diabetes-related features, showcasing how quickly one can build and evaluate multiple classification algorithms with “low code”, visualise their performance, and predict outcomes on new data. Manjunath, at the time of writing, is a principal AI engineer at msg global solutions, working on graph databases, generative AI solutions, and more.
Srihari Thyagarajan’s talk on marimo, an open-source, reactive Python notebook, showcased an innovative notebook format that is, in marimo’s words, “a next-generation reactive notebook, stored as Git-friendly, reproducible, deployable as a script, shareable as an app”, that enhances the interactivity and usability of Python notebooks as compared to the standard Jupyter notebook format. We were thrilled to see the enthusiasm for marimo, especially its potential for AI-related integrations and scientific use cases, and its ability to facilitate reproducible in-browser research via WebAssembly. The discussion also highlighted exciting opportunities for community contributions and integration projects with marimo, similar to how Jupyter has fostered a thriving ecosystem of extensions and plugins. As Srihari mentioned, he enjoyed his time presenting to the community and getting feedback as an early contributor, ambassador, and intern at the venture, working on topics such as interactive docs, integrations, outreach, and more.
Mohammad Razak’s lightning talk on causal machine learning and the pgmpy library provided a concise introduction to the concepts of causality in machine learning, and how pgmpy can be used to model and infer causal relationships in data. He discussed the importance of understanding causality for making informed decisions based on data (i.e., the “correlation does not imply causation” adage), and how pgmpy provides tools for building probabilistic graphical models that can capture these relationships. His talk sparked interest in the community about the applications of causal inference in various domains, including healthcare, economics, and social sciences. Mohammad Razak, at the time of writing, is developing a Rust backend for pgmpy as part of an European Summer of Code (ESoC) project, and we look forward to seeing the progress on this front.
Varuni H K’s lightning talk on subgraph matching addressed the challenges of finding smaller graphs within larger ones, a problem that is becoming increasingly important in fields such as social network analysis, bioinformatics, and recommendation systems. She introduced the concept of subgraph isomorphism and discussed various algorithms and techniques for efficient subgraph matching, including graph neural networks. Her talk generated a lively discussion among participants interested in network analyses about the potential applications of these techniques and the need for further research in this area. Varuni, at the time of writing, is a software engineer at CouchBase, working on distributed databases and vector search.
In between the talks, discussions engaged participants on various topics, including the future of SciPy India, the licensing and governance of libraries used in scientific Python projects, questions on venture capital funding for open source projects and its implications on sustainability, and several other topics.
Aftermath#
The community call concluded with surveys to gather feedback from attendees, and gauge if there is interest in organising such events in person at intervals throughout the year, in addition to our regular online calls. We are also exploring if there is interest in the community for the next SciPy India conference in late 2025 or early 2026, through another survey that was shared during the call. The survey will remain open, and we encourage you to fill it out. We also received feedback on the content and overall format of the call, as well as constructive suggestions on topics for future calls, which we will take into consideration and incorporate into our planning for upcoming events. Many attendees expressed interest in volunteering and contributing to the community, which is heartening for us to see!
Livestream#
Our community call was livestreamed on YouTube for those who could not attend the live session. The recording is available for viewing at your convenience.
In a nutshell#
Thank you to everyone who participated and contributed to the success of our first community call! Especially, we thank our speakers for their presentations, and the discussions that followed. In addition, we are grateful to the FOSS United platform and its representatives for providing us with the necessary infrastructure to host our event online and manage registrations.
We look forward to seeing you at upcoming community calls and events! To stay updated on the same, please join our Zulip chat and follow us on our social media channels. And to you, the reader, thank you for reading!