InternMatch - AI-Powered Internship Recommendation Engine

An intelligent internship matching system that uses Machine Learning to recommend the best PM internships based on student profiles. Built with Streamlit and powered by content-based filtering using cosine similarity.

🎯 Features

1. Student Profile Input

   • Name, skills, and location preferences
   • Internship type selection
   • Easy-to-use sidebar interface

2. Sample Internship Dataset

   • 10 curated PM internship opportunities
   • Diverse companies and locations
   • Detailed job descriptions and requirements

3. ML-Based Matching Algorithm

   • Content-based filtering using TF-IDF vectorization
   • Cosine similarity for matching student profiles to internships
   • Match scores (0-100%) for each recommendation

4. Top 3 Recommendations Display

   • Visual match score indicators (🟢 70%+, 🟡 50-70%, 🔴 <50%)
   • Detailed internship information
   • Company, location, duration, and required skills

5. Feedback System

   • Accept/Reject buttons for each recommendation
   • Feedback history tracking
   • Downloadable JSON feedback data
   • Analytics dashboard with statistics

🚀 Quick Start

Prerequisites

• Python 3.11 (already installed)
• Required packages (already installed):
  • streamlit
  • pandas
  • scikit-learn
  • numpy

Running the Application

Option 1: Using the run script (Recommended)

./run_streamlit.sh

Option 2: Direct command

streamlit run app.py --server.port 5000 --server.address 0.0.0.0

Option 3: Default Streamlit port

streamlit run app.py

The app will be available at:

• Local: http://localhost:5000 (or http://localhost:8501 for default)
• Replit: Will be accessible via the Replit preview URL

📁 Project Structure

.
├── app.py                      # Main Streamlit application
├── internship_data.py          # Sample internship dataset
├── recommendation_engine.py    # ML recommendation engine (cosine similarity)
├── run_streamlit.sh           # Convenience script to run the app
└── README.md                   # This file

🧠 How It Works

1. Data Loading

The system loads a curated dataset of 10 PM internship opportunities with:

• Job titles and descriptions
• Required skills
• Location and duration
• Company information

2. Feature Extraction

• Combines internship description, skills, and location into a single content string
• Uses TF-IDF (Term Frequency-Inverse Document Frequency) to vectorize text
• Creates numerical representations of internships and student profiles

3. Similarity Calculation

• Transforms student profile (skills + location) into TF-IDF vector
• Calculates cosine similarity between student and all internships
• Cosine similarity measures the angle between vectors (0-1 scale)

4. Recommendation Generation

• Ranks internships by similarity score
• Returns top 3 matches with percentage scores (0-100%)
• Higher scores indicate better matches

💡 Usage Example

1. Enter Your Profile:

   • Name: "Sarah Johnson"
   • Skills: "product strategy, analytics, SQL, user research"
   • Location: "Remote"

2. Click "Get Recommendations"

3. View Results: See your top 3 matches with scores like:

   • 🟢 Product Management Intern @ TechCorp - 92.5%
   • 🟢 Growth Product Intern @ GrowthLabs - 85.3%
   • 🟡 Product Strategy Intern @ ConsultCo - 78.9%

4. Provide Feedback: Click ✅ Accept or ❌ Reject for each recommendation

5. Review Feedback: View feedback history and download data

🔧 Technical Details

ML Algorithm

• Type: Content-Based Filtering
• Vectorization: TF-IDF (sklearn.feature_extraction.text.TfidfVectorizer)
• Similarity Metric: Cosine Similarity (sklearn.metrics.pairwise.cosine_similarity)
• Features Used: Job description, required skills, location

Why Cosine Similarity?

• Measures semantic similarity between text documents
• Scale-invariant (works well with different text lengths)
• Values range from 0 (no similarity) to 1 (identical)
• Industry standard for content-based recommendation systems

📊 Feedback System

The app tracks user feedback to help improve recommendations:

• Accept/Reject: Simple binary feedback
• Timestamp: When feedback was given
• Match Score: AI-calculated match percentage
• Export: Download feedback as JSON for analysis

🎨 UI Features

• Clean, modern Streamlit interface
• Responsive sidebar for profile input
• Color-coded match scores (green/yellow/red)
• Expandable feedback history
• Real-time statistics dashboard

🔮 Future Enhancements

Potential improvements for the next phase:

• Collaborative filtering using historical feedback data
• Hybrid recommendation (content + collaborative)
• Resume parsing with NLP
• Skill gap analysis
• Email notifications for new matches
• A/B testing for algorithm improvements

📝 License

This project is for educational and demonstration purposes.

---

Built with ❤️ using Streamlit, scikit-learn, and Python "# INTER_MATCH_AI_MODEL"