{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/6540fad032f34d23a31af6bae7b683ce\" frameborder=\"0\" width=\"1728\" height=\"1296\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1296,"width":1728,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1296,"thumbnail_width":1728,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/6540fad032f34d23a31af6bae7b683ce-5c88005da897056c.gif","duration":285.785,"title":"Building a Music Recommender with Guardrails 🎧","description":"I built an applied AI system for a content based music recommender that takes user preferences like genre, mood, and energy, then ranks songs with a scoring function. I extended a basic recommender with evaluation, confidence scoring, and guardrails for reliability and interpretability. In recommender logic, each song is scored on match, mood match, and continuous energy similarity for more nuanced ranking. I ran profiles like High energy pop and Show Lofi, showing high scoring, normalized confidence, and edge case fallbacks. I also added an Evaluator to measure average score and genre diversity across profiles."}