Resume Chat

Personal Inquiry Agent

Why I Built This

Recruiters and HR professionals often want to get quick answers to basic questions:\

• “Can you walk me through your background?”
• “What kind of projects have you worked on?”
• “Are you open to relocation or remote?”

Answering these repeatedly can feel like a loop. So, I decided to build Resume Chat: an intelligent, interactive agent that represents me online, answers career-related questions in real time, and even learns from what it cannot answer yet.

You can try it live on my website: 👉 evotianusx.github.io/chat

---

Setup

How It Works

Under the hood, Resume Chat is a lightweight Python application that combines OpenAI’s models, document parsing, and a Gradio-powered interface. It’s essentially a “virtual me” that HR can interact with---kind of like an AI-powered cover letter that talks back.

Key features include:

• Context-Aware Responses — It loads a detailed summary and my LinkedIn export to generate personalized answers.
• Feedback Loop — Unanswered questions are logged for me to refine later.
• Real-Time Alerts — Via Pushover, I get notified whenever a recruiter leaves their details or asks something new
• Interactive Web UI — A simple, clean Gradio chat interface that embeds neatly into my personal site.

---

Technical Setup

1. Environment Variables

All secrets are stored in a .env file (and ignored by Git):

OPEN_AI_KEY=your_openai_api_key
PUSHOVER_TOKEN=your_pushover_application_token
PUSHOVER_USER=your_pushover_user_key

2. Context Files

The agent reads my background from local files:

│   app.py
│   README.md
│   requirements.txt
│
└───me
        linkedin.pdf
        projects.yaml
        summary.txt

Installation

1. Clone
2. uv pip install requirements.txt
3. uv run app.py

This starts a Gradio web server at http://127.0.0.1:7860.

---

Deployment to Hugging Face Spaces

Since this is a static site add-on, I needed something free, simple, and reliable. Hugging Face Spaces with Gradio turned out perfect.

Steps to deploy:

1. Install Gradio CLI
   bash pip install --upgrade gradio

2. Authenticate
   bash huggingface-cli login

3. Deploy
   bash gradio deploy app.py (Replace with your filename if different, e.g. main.py.)

4. Secrets
   Save your keys in secrets.txt (never commit!) and Hugging Face will handle encryption.

Once deployed, it’s live at:

👉 https://huggingface.co/spaces/<username>/<space-name>

Mine lives here: evotianusx/evo_resume_chat.

Embedding it into my website was as simple as dropping an iframe.

---

Lessons Learned

• Recruiter questions tend to repeat---a conversational agent really does cut down redundancy.\
• Handling context sources cleanly (summary + LinkedIn + projects) is trickier than I expected. Prompt balance matters.\
• Deployment on Spaces is smooth, but environment variables must be handled carefully.\
• The feedback loop (logging unanswered questions) makes the system feel like it “improves” over time.

---

🚀 For me, Resume Chat is more than a side project---it’s a small experiment in making the job hunt less repetitive and more interactive.