Hi! I'm Rahul.
I'm building multimodal AI systems that help us understand visual storytelling.
Search and discovery engine for film and TV. We're building narrative intelligence to understand stories at scale. Our AI understands elements of craft, plot, emotional subtext, etc. enabling new ways of interacting with content.
Presented CinemaNet, a jointly trained multi-task classification model that codifies a comprehensive taxonomy of cinematography, at the inaugral Creative Video Editing and Understanding workshop.
Flagship interactive retail experience for HBO Max & AT&T, in collaboration with HUSH, where users could interact with parts of the HBO archive in real-time with gestures and words.
Collaborative project with IYOIYO studio / Google & YouTube. I built custom vision models that helped power the visual intelligence behind the experience.
Blog post introducing the basics of visual language in cinema, and a custom model and dataset built to recognise different types of framing.