RAG Document Assistant

Overview

Built a retrieval-augmented chatbot for querying documents with source attribution.

Responsibilities

• Implemented document ingestion pipeline
• Built vector search system
• Designed answer + citation output

Approach

• Embedding-based retrieval
• Context injection into prompts
• Source-grounded answering

Architecture

flowchart LR
    A[User Query] --> B[Embedding]
    B --> C[Vector DB]
    C --> D[Relevant Docs]
    D --> E[LLM]
    E --> F[Answer + Sources]

Tech

OpenAI · Vector DB

Impact

• Improved answer accuracy with grounding
• Enabled explainable AI responses
• Reduced hallucination risk