Talk
Software testing is a cornerstone of reliable systems, yet it is often overlooked in data science and machine learning projects. Unlike traditional applications, data workflows introduce unique challenges such as data variability, silent failures, and reproducibility issues.
This talk focuses on applying practical testing techniques using pytest in the context of Python-based data and ML systems. Drawing from real-world scenarios, we will explore how to design and implement effective tests for data transformations, pipelines, and model outputs.
We begin by examining common pitfalls in untested data workflows and why traditional testing approaches often fall short. We then introduce core pytest features such as fixtures and parameterized tests, demonstrating how they can be used to create robust and maintainable test suites.
The session also covers strategies for testing different layers of a system, including unit tests for functions, integration tests for pipelines, and validation techniques for machine learning models. Additionally, we will touch on testing in the era of AI-assisted development, highlighting how automated tests help ensure correctness and reliability.
By the end of the talk, attendees will have a clear understanding of how to incorporate testing into their data workflows, improve code quality, and reduce the risk of failures in production systems.
About the Speaker
Hi, I’m Sneha, a software engineer and QA specialist with 3 years of experience. I currently work at HighLevel, where I help teams ship reliable products by making sure web, mobile, and backend systems work smoothly for real users. From manual testing to automation, my goal is simple catch issues early so users never feel the pain.
Before HighLevel, I worked at Swiggy, CloudDefense.AI, Morgan Stanley, and Wingify. These roles exposed me to fast moving products, cloud security, and large scale systems. They shaped how I think about quality not just as finding bugs, but as deeply understanding users, edge cases, and long term product health.
Outside my day job, I’m a LinkedIn content creator with over 31,000 followers. I love breaking down complex tech topics into simple, practical ideas around software careers, AI use cases, and everyday problem solving. Teaching and learning in public is something I genuinely enjoy.
I’m also an active speaker in the global Python community. I’ve spoken at PyCon APAC 2025 and PyCon US 2025, and my talks have been selected for PyCon Sweden, PyCon Spain, PyCon Colombia, and the Swiss Python Summit. Speaking allows me to share real world lessons and learn from developers across the globe.
At heart, I’m curious, detail oriented, and driven by impact. I care deeply about building technology that works well, scales gracefully, and genuinely makes life easier for people.