I’m Purvanshi Mehta, Founder and CTO of Lica, where we’re building foundational multimodal models to power graphic design and creative workflows at scale. With over seven years of experience in AI, I’ve led research in multimodal learning since 2018, including early work on interpretability, and have presented at top conferences like NeurIPS, IJCAI, AAAI, and IJCNLP. Previously, I worked on the early CoPilot teams at Microsoft that brought large language models into production across Office products, serving over 400 million users. At Lica, I lead a team of researchers and engineers reimagining how AI can become the creative brain of a brand. As a speaker at the AI Engineering Summit, I’m excited to share why design remains a blind spot for most models—and how we’re solving it.