HS.
All projectsEngineering case study

INDEPENDENT FULL-STACK PROJECT

ShelfSense — book discovery with cited retrieval

Django/Next.js book discovery combining ingestion, cached Gemini insights, and source-linked retrieval.

Himanshu SharmaCase study updated

What I built

A book-discovery application that combines Selenium ingestion, MySQL metadata, Gemini-generated insights, and ChromaDB retrieval with a Next.js frontend. The retrieval flow uses overlapping chunks and source-linked results. Content-hash caching reuses generated insights, while persisted Q&A history keeps previous interactions available.

Scope

The retrieval index covers book descriptions and metadata. I implemented this as an assignment project with backend/frontend interaction tests. Stack: Django, Next.js, MySQL, Gemini, ChromaDB, Selenium. View the source code
All projects