video1.0<iframe src="https://www.loom.com/embed/f374780fbd334954851f8425f9d2434e" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/f374780fbd334954851f8425f9d2434e-b40c387bd6fe2ca0.gif378.483Today I demoed my final AI engineering project, eco-mind, a hybrid music recommender. It combines a content similarity score from audio features and a label based preference score from gender, mood, and energy, then applies an agentic policy layer using intent text and session feedback like likes and skips. I ran three inputs to show how outputs change, including reliability guardrails where conflicting inputs still return a stable ranked list. I also covered evaluation results with 22 passing tests out of 23. No action was explicitly requested from viewers.