Team Surya

Llama-Persona is an AI-powered contextual advertising prototype that explores how digital ads can become more relevant to a user’s current activity instead of relying primarily on static profiles and tracking cookies. Traditional advertising often relies on broad audience segments or historical tracking, which can lead to irrelevant ads and limited transparency about why an ad is shown. Llama-Persona addresses this by analyzing activity context, identifying potential commercial intent with Llama-powered AI, and generating context-aware ad concepts. Privacy is a core part of the prototype. Activity telemetry is processed locally, and the system uses structured activity signals rather than sending raw browsing content to an advertising platform. The application includes a Privacy Center with controls to pause or manage activity collection, clear stored activity data, and inspect how data is handled. It also provides explanations for why an ad was selected, helping users understand the connection between their activity context and generated recommendations. The project demonstrates a more contextual, explainable, and privacy-conscious approach to advertising through an activity timeline, behavioral analytics, intent prediction, ad generation, privacy controls, and an evaluation dashboard.