video1.0<iframe src="https://www.loom.com/embed/2d0480d1002a49cc99760d865e668688" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/2d0480d1002a49cc99760d865e668688-edc738619e527fe9.gif388.864Hi, I’m Megashree. For Mom’s World, I built an AI shopping assistant that converts natural language queries like gift for newborn under 150 dirhams into structured product recommendations in English and Arabic. I designed a system that filters relevant products, sends them to an LLM, validates the JSON output with retries and a rule based fallback for API rate limits, and handles unclear inputs with safe responses to avoid hallucinations. I also created an evaluation pipeline with Normal and EdgeCase test cases to measure success on structured output and relevance. No specific action was requested from viewers.