VectorRAGNuxt
Learning vector search and retrieval-augmented generation with Nuxt and MongoDB, using real car data.
Nuxt, Node, MongoDB Vector Search, OpenAI API
A hands-on study of vector databases and RAG. You describe a car in plain language and get back the closest real matches.
What’s in it
- A data scraper pulls model data from the NHTSA vPIC API, filterable by year range and make.
- MongoDB as the vector store. Scraped records are embedded and searched by similarity.
- A Nuxt front end takes the prompt and shows the results.
Status
Search works end to end: scraped models are seeded into MongoDB, and a plain-language description comes back as the closest matches with a score. The results are still rough (a “medium sedan” search turns up an SL-Class roadster), so tuning what gets embedded is next.