Vijay believes nothing is truly that hard — it’s just that some things are less easy.
He is a machine learning software engineer currently working to build Kooper AI (https://hellokooper.com), a platform reimagining reading and creative expression for kids through generative AI. His research interests lie at the intersection of mathematics, computer science and statistics. With a background in AI/ML product development and building human-centered technologies, Vijay focuses on designing AI systems that balance innovation, safety, and real-world impact — helping the next generation become not just users of AI, but confident creators with it. Outside of work, Vijay is passionate about Indian (Carnatic) classical music.
Building Safe, Creative AI Experiences for Kids: Real-World Lessons from Kooper AI
What happens when you put generative AI in the hands of the most imaginative — and unpredictable — users: kids? At Kooper AI, we set out to create a platform where kids don’t just read stories, but create, remix, and share them using AI — safely and creatively. One of our biggest challenges was balancing creative freedom with responsible safety controls: how to design systems that encourage exploration without opening the door to inappropriate or unsafe content.
In this talk, I’ll share practical lessons from taking a multimodal AI system into production for young users — including lightweight alignment techniques for steering outputs, prompt and template design strategies to nurture creativity within safe boundaries, and methods for managing unstructured content like user-generated text and images. I’ll also cover how we connected AI behaviors directly to product metrics like reading engagement, creativity scores, and safe social interaction.
Whether you’re building for kids or adults, these frameworks offer actionable ways to create AI systems that empower users — without compromising trust.