OlajCodes AI: The Autonomous Portfolio Agent

"Don't just read my CV. Interview my AI agent."

This is not a static portfolio website. It is an autonomous RAG (Retrieval-Augmented Generation) Agent that represents me professionally. It ingests my entire digital footprint (GitHub code, LinkedIn profile, Technical Articles), indexes it into a Vector Database, and uses it to answer recruiter questions and generate tailored documents in real-time.

🔗 Live Demo: Personified RAG

🚀 Key Features

1. 🧠 Evidence-Based RAG Chat

Unlike generic chatbots, this agent cannot hallucinate.

• Strict Grounding: Every claim is cross-referenced with my actual data.
• Citations: Responses include clickable citations (e.g., [Source: main.py], [Source: Profile.pdf]) proving I actually did the work.

2. 📄 Tailored Document Generation

The Recruiter Value Driver:

• Input: Paste any Job Description (JD).
• Process: The AI maps my skills to the JD requirements.
• Output: Instantly downloads a Tailored CV (.docx) and Persuasive Cover Letter (.docx) optimized for that specific role.

3. 🛡️ Intelligent Guardrails ("The Gatekeeper")

The agent acts as a professional filter:

• Relevance Check: If a user pastes a JD for a non-tech role (e.g., "Nurse" or "Chef"), the AI refuses to generate documents, explaining that the role does not align with my expertise.
• Privacy: It is programmed to withhold sensitive personal data (phone number, home address) unless specifically authorized.

4. 🎨 Modern Dual-Theme UI

• Dynamic Theming: Seamless toggle between a crisp Corporate Light Mode (White/Green) and an immersive Developer Dark Mode (Gray/Emerald).
• Responsive: Fully mobile-optimized sidebar and chat interface.

---

🏗️ System Architecture

The system follows a standard RAG pipeline optimized for accuracy:

graph LR
    A[Data Sources] -->|Ingest| B(LangChain Loader)
    B -->|Chunking| C(Text Splitter)
    C -->|Embed| D[OpenAI Embeddings]
    D -->|Store| E[(ChromaDB Vector Store)]
    
    User[User Query/JD] -->|Search| E
    E -->|Retrieve Context| F{GPT-4o Logic}
    F -->|Answer| Chat[Chat Response]
    F -->|Generate| Doc[Tailored .docx]

Loading

1. Ingestion: Scrapes my GitHub repos, parses my LinkedIn PDF, and reads local Markdown notes.

2. Embedding: Uses text-embedding-3-small to convert text into vector space.

3. Retrieval: ChromaDB performs semantic search to find relevant experience.

4. Generation: GPT-4o synthesizes the answer or drafts the document using python-docx.

🛠️ Tech Stack

Backend (Python)

• FastAPI: High-performance async API framework.

• LangChain: Framework for RAG logic and chain orchestration.

• ChromaDB: Local vector database for context storage.

• OpenAI API: LLM (GPT-4o) and Embeddings.

• Python-docx: Programmatic document creation.

Frontend (React)

• Vite + React: Fast, modern frontend build tool.

• Tailwind CSS: Utility-first styling for the custom "Green/Emerald" theme.

• Lucide React: Beautiful, consistent iconography.

• Fetch API: For streaming responses and file downloads.

⚡ Quick Start

Prerequisites

• Python 3.10+

• Node.js 18+

• OpenAI API Key

1. Clone & Setup Backend

git clone [https://github.com/Olajcodes/personified-rag.git](https://github.com/Olajcodes/personified-rag.git)
cd personified-rag/backend

2. Create Virtual Env

python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows

3. Install Dependencies

pip install -r requirements.txt

4. Set API Key

• Create a .env file and add: OPENAI_API_KEY=sk-proj-...

5. Ingest Data Before running the server, you must build the "Brain" (Vector DB).

python ingest.py

You should see: "Success! Database created."

Run Servers Backend:

uvicorn main:app --reload

Running on http://localhost:8000

Frontend:

cd ../frontend
npm install
npm run dev

Running on http://localhost:5173

The Screenshot of the Landing Page.

The Screenshot of the Chat Interface.

📄 License

This project is proprietary and confidential. Copyright © 2026 OlajCodes.