Show HN: I built Search/Recsys that learns from feedback

Category: ai-ml

Tags: vector-search, recommendation-system, llm-as-a-judge, feedback-loop, embeddings

Score: 5.7/10 (Innovation: 6, Technical: 6, Documentation: 2, Utility: 5)

DNA is a search/recsys API that stores one embedding per item and improves matches in real time using feedback scores (e.g., from LLMs or users). It claims to reduce LLM judgments by 7x compared to scoring down the embedding list. The project is an early-stage Show HN post with no documentation, license, or source code available.

Target audience: backend devs, data engineers, ML engineers

Repository: https://dna.hashgrid.ai/

View on Hacker News