Building Beyond the Day Job: Divya’s Data Science Growth on Zindi
Divya Challa (LinkedIn) holds a Master’s degree in Computer Science and works as a software engineer. While her day job gives her plenty of exposure to how ML systems are built and used, she doesn’t often get the chance to build models herself. She saw the International Women’s Day Challenge as a great way to step outside her day-to-day responsibilities and practice those skills while staying updated on the latest machine learning techniques. She ranked first on the challenge, taking home her first Zindi win.
“What made it even more special was that the challenge focused on women. As a woman in tech, I found that especially inspiring and encouraging,” Divya shared.
Divya approached the challenge methodically. She started with exploratory data analysis, checking distributions, missing values, and feature correlations. From there, she tested several baseline models like Random Forest, Decision Trees, and XGBoost, before moving on to more sophisticated ensemble methods. Eventually, a weighted blended model using algorithms like LightGBM and XGBoost gave her the best results. To refine her solution, she used SHAP values for model explainability and feature selection.
“The challenge helped strengthen my understanding of end-to-end ML workflows, from data prep and feature selection to ensemble modeling and explainability,” she says in a recent interview. “Zindi allows me to apply my data science skills to real-world problems. I’ve had a lot of exposure to how ML systems are used on the job, but I don’t always get to build models myself, so challenges like these help me stay hands-on and keep learning.”
Beyond the win, Divya is motivated by the way machine learning is evolving. She’s especially interested in representation learning, where models can learn patterns directly from raw data. It’s something she finds exciting in fields like NLP and computer vision, where systems are starting to pick up on relationships and features in much the same way people do. Self-supervised learning is another area she’s drawn to, where models can teach themselves from unlabeled data, an approach that brings us closer to more human-like learning.
Divya enjoys reading, traveling, hiking, and experimenting with new recipes in the kitchen.
Looking ahead, she wants to keep applying her skills through more projects and competitions, and encourages others to do the same. “Focus on the basics, practice consistently, and learn by doing. Challenges are a great way to grow and stay motivated.”
