I built a video search engine that actually understands what's being said and shown. Try it and provide Feedback.

What I want from you

Be brutal. Specifically:

• 
  Search quality: Did it answer your question? Were the timestamps right? Did it hallucinate?


• 
  UI/UX: Anything confusing? Anything ugly?


• 
  Speed: How long did queries take? Was it annoying?


• 
  Broken stuff: Errors, weird results, blank screens. Screenshot them if you can.

Drop feedback in the comments, or open an issue on GitHub. Both work.

Why I'm posting this

I'm an AI engineer and this is a portfolio project. I vibecoded most of it with Claude and then spent my time on the parts that matter: tuning retrieval thresholds, running eval sets, and figuring out why hybrid search with BM25 kept returning garbage until I got the alpha weighting right.

But portfolio projects have a blind spot: I've been staring at this thing for days straight. I need fresh eyes.

If you work on RAG, search, or video ML, I'd especially love your take on the retrieval pipeline. The eval harness has 145 golden queries but I know there are gaps.

Thanks for reading. Go break my stuff.