Resume Chatbot

Ask any questions about my professional experience, skills, or background. Try both modes - Basic and RAG (Retrieval-Augmented Generation) - to see how different AI approaches affect responses.

Resume Chat

Hi there! I'm an AI assistant who can answer questions about David Larrimore's professional experience, skills, and background. You can switch between Basic and RAG modes using the toggle below. What would you like to know?

Basic Mode

Basic Mode: Uses the entire resume as context for every question.

RAG Mode: Uses Pinecone's embedding model to retrieve only the most relevant parts of the resume based on your question, potentially giving more focused answers.

About This Project

How It Was Built

This Resume Chatbot was built using Node, Next.js, Tailwind CSS, Pinecone (Vector Store) and Anthropic's Claude AI. The application features two different operational modes that demonstrate different approaches to AI-powered chat:

  • Basic Mode: Sends my entire resume as context to Claude with each user query, allowing the AI to reference all information.
  • RAG Mode: Uses Pinecone vector database to store embeddings of my resume chunks. When you ask a question, it retrieves only the most relevant sections and sends those as context to Claude.

Technical Implementation

  • Frontend: React components with client-side state management using the useState hook.
  • API Routes: Next.js API routes to handle communication with Anthropic's Claude API.
  • RAG Architecture: Resume data is chunked, embedded, and stored in Pinecone's vector database for semantic search.
  • Response Rendering: ReactMarkdown for formatting Claude's responses with proper typography and styling.

Claude AI Prompt Details

The system prompt sent to Claude varies based on the mode selected:

Basic Mode Prompt Structure:

  • Defines the assistant's role as a helpful resource for questions about David's professional background
  • Instructs the AI to use only information from the provided resume
  • Sets expectations for response style: concise, friendly, and professional
  • Includes formatting guidelines for better readability
  • Provides the complete resume content as context for every question

RAG Mode Prompt Structure:

  • Sets a more conversational, enthusiastic tone for the assistant
  • Includes basic biographical information about David
  • Provides a high-level summary of his career progression
  • Incorporates contextually relevant resume chunks retrieved from the vector database based on your question
  • Features specific guidelines for answer length, accuracy, and relevance
  • Includes instructions for handling questions outside the scope of available information
  • Emphasizes factual responses based only on the information provided

Why Explore Both Modes?

Comparing the responses from Basic and RAG modes demonstrates the tradeoffs in AI systems. RAG mode can provide more precise answers by focusing only on relevant information, which can reduce hallucinations and improve answer quality. However, it may sometimes miss context from other sections that could be valuable. Basic mode has all information available but might include irrelevant details or get distracted by unrelated information.