Selected work / 04
ChatWithPDF
- Focus
- AI document search
- Role
- Full Stack Developer
- Year
- 2025
CHATWITHPDF / DOCUMENT INTELLIGENCE↗ 2025
Beyond
the page.
YOUR DOCUMENTS HAVE ANSWERS.AK®
Original conceptual project cover. This artwork illustrates the project theme.
The problem
Long PDF documents can make it difficult to find a specific answer. The product lets people upload a PDF and explore its contents through natural-language questions.
The approach
A retrieval-augmented generation workflow combines Gemini embeddings with Pinecone semantic search. PostgreSQL and Drizzle ORM support conversation history alongside the Next.js interface.
Next.jsTypeScriptGeminiPineconePostgreSQLDrizzle ORM
How the pieces connect
A simplified view of the main layers and responsibilities.
PDF contentDocument input
Gemini embeddingsSemantic representation
Pinecone retrievalRelevant context
Contextual answersConversation history
What the work delivers
The result is a document interaction platform built around semantic retrieval and contextual answers.
Discuss this project ↗NEXT PROJECT / 01
↗