Interactive Resume Project Overview

The Interactive Resume AI assistant is a custom-built, AI-powered tool embedded on charlesbartlett.net that allows visitors — particularly hiring managers and recruiters — to ask natural language questions about Charles Bartlett’s professional background and receive accurate, source-referenced answers instantly.

Rather than reading through multiple pages on the site, a visitor can simply ask a question like “What experience does Charles have in sales enablement?” and receive a concise, relevant answer drawn directly from the published content on the site.

What It Does

The assistant allows visitors to:

  • Ask open-ended questions about Charles’s professional experience, skills, and background
  • Receive answers written in clear, professional third-person prose
  • See which pages on the site the answer was sourced from, with clickable links to read more
  • Have a short back-and-forth conversation, with the assistant remembering recent context

The assistant only answers questions using content that has been explicitly published on the site. It will not speculate, fabricate, or draw on outside knowledge. If a question cannot be answered from the available content, it says so clearly.

How It Works

The assistant uses a modern AI technique called Retrieval-Augmented Generation (RAG). In plain terms:

  1. Content is indexed. When a page or post is published and marked for inclusion, the assistant reads and stores its meaning in a way that allows it to be searched by concept — not just keyword.
  2. Questions are understood by meaning. When a visitor asks a question, the assistant finds the most relevant passages from the indexed content — even if the exact words don’t match.
  3. Answers are generated, not retrieved. A large language model reads the relevant passages and writes a natural, concise answer — always grounded in what the site actually says.

The assistant is embedded directly on the website with no third-party chat platform required. It appears as a floating chat button on every page, or as an inline widget on specific pages.

Technology Stack

  • Cloud Infrastructure: Amazon Web Services (AWS) — serverless, scales automatically, no servers to manage
  • Vector Search: Pinecone — industry-standard vector database used for semantic search
  • AI Embeddings: Voyage AI — converts text into mathematical representations that capture meaning
  • AI Generation: Anthropic Claude — the large language model that writes the final answers
  • CMS Integration: Custom WordPress plugin — manages which content is indexed and handles all communication between the website and the AI backend

Content Control

Not all content on the site is automatically included. Each page or post must be explicitly opted in by checking a box in the WordPress editor. An admin column in the WordPress dashboard shows at a glance which content is included in the index, and a filter makes it easy to review. If a page is unchecked and saved, its content is automatically removed from the assistant’s knowledge base.

Privacy and Safety

  • The assistant does not store personally identifiable visitor information beyond an anonymized IP address used for rate limiting
  • Questions and answers are logged for quality review and improvement purposes
  • Rate limiting prevents abuse — each visitor is limited to a set number of questions per hour
  • All communication between the website and the AI backend is cryptographically signed to prevent tampering
  • The assistant will not answer questions outside the scope of the published site content

Current Status

The assistant is fully deployed and operational on charlesbartlett.net. Content is actively being indexed and the assistant responds accurately to questions about Charles’s professional experience, roles, skills, and accomplishments.

Potential Future Enhancements

  • Index additional pages and content types as the site grows
  • Add suggested starter questions to help visitors get started
  • Build an analytics dashboard to review the most common questions asked
  • Fine-tune response style and formatting based on visitor feedback