• 

• 

• 

• 

• 

  architecture

📄 Resume Screener Bot 🤖

✨ Inspiration

Job seekers often struggle to understand how their resumes are perceived by recruiters or AI screening tools. I wanted to build a simple, intelligent tool that could:

• Analyze a user's resume
  
• Allow natural-language questions
  
• Provide smart feedback using a local AI model (Mistral via Ollama)

I was inspired by the idea of helping students and freshers prepare better for job applications using accessible, open-source tools.

🛠 How I Built It

• Frontend: Built with Gradio for an interactive, easy-to-use interface
  
• Backend: Python + Ollama running the mistral model locally
  
• Resume Parsing: Used PyMuPDF (fitz) to extract clean text from PDF resumes
  
• AI Integration: Prompted Mistral to provide contextual answers based on resume text
  
• Deployment: UI hosted on Render, backend tunneled with ngrok

💡 What I Learned

• How to run local LLMs using Ollama
  
• Building no-code interfaces with Gradio
  
• Prompt engineering for resume review tasks
  
• Combining cloud UI with local AI inference via tunnels
  

🚧 Challenges Faced

• ❌ Ollama can’t be deployed on Render, so I used ngrok to expose my local API
  
• 🧾 Handling unstructured and large resumes while preserving formatting
  
• ⚙️ Managing dependency issues (like PyMuPDF on Windows)

🚀 Future Enhancements

• 🎤 Add voice input and TTS replies
  
• 🧠 Job-matching suggestions based on resume content
  
• 🔁 Switch between Ollama and OpenAI automatically

🙌 Credits

Built with ❤️ using:

• Python
  
• Gradio
  
• Ollama
  
• PyMuPDF
  
• Render
  
• Ngrok
  

🎥 Demo

Watch how the Resume Screener Bot works and answers real resume questions!
Tag: #Devpost

Built With

• fitz
• gradio
• mistral
• ngrok
• ollama
• pymupdf
• python
• render