🎯 Inter-Sight - Intelligent Talent Matching Platform

Status: MVP for Hack-Nation Global AI Hackathon (Nov 8-9, 2025)

What is Inter-Sight?

Inter-Sight is a feedback-first talent intelligence platform that uses AI to:

• ✅ Extract soft skills intelligently (semantic, hybrid approach)
• ✅ Match candidates to company culture and values
• ✅ Detect red flags with context and reasoning
• ✅ Generate personalized, actionable feedback for candidates
• ✅ Provide beautiful Tinder-like candidate ranking interface
• ✅ Process batch CVs with comprehensive analysis

Why Inter-Sight?

The Problem: Hiring today is broken.

• Recruiters drown in CVs with no context
• Candidates get rejected without feedback
• Companies miss hidden talent
• Culture mismatches happen months after hire

The Solution: Intelligent feedback + soft skills + culture alignment

Quick Start

Prerequisites

• Python 3.10+
• Mistral API key (free tier: https://console.mistral.ai)

Setup

# Clone or extract this repository
cd inter-sight

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Create .env file
cp .env.example .env
# Edit .env and add your MISTRAL_API_KEY

Run Application

streamlit run ui/streamlit_app.py

Open browser to: http://localhost:8501

Project Structure

inter-sight/
├── core/                          # Core modules
│   ├── config.py                 # Configuration management
│   ├── llm_provider.py           # LLM abstraction layer
│   └── constants.py              # Constants & schemas
│
├── processors/                    # Processing pipeline
│   ├── company_profiler.py       # Company profile builder
│   ├── cv_parser.py              # CV parsing & structure extraction
│   ├── skill_extractor.py        # Skill extraction (hybrid)
│   ├── culture_analyzer.py       # Culture value matching
│   ├── red_flag_detector.py      # Red flag detection
│   ├── matcher.py                # Multi-layer matching
│   └── feedback_generator.py     # Feedback generation
│
├── ui/                            # Streamlit UI
│   ├── streamlit_app.py          # Main application
│   └── components/               # UI components
│
├── data/                          # Data files
│   ├── sample_cvs/               # Sample CV data
│   └── sample_companies/         # Sample company profiles
│
├── utils/                         # Utility functions
│   ├── logger.py                 # Logging setup
│   ├── validators.py             # Input validation
│   └── formatters.py             # Output formatting
│
├── docs/                          # Documentation
├── requirements.txt              # Python dependencies
├── .env.example                  # Environment template
└── README.md                     # This file

Features

1. LLM Abstraction Layer

Support for multiple LLM providers:

• Mistral (default, free tier)
• Claude (Anthropic)
• OpenAI (GPT)
• Local Llama (Ollama)

Just change LLM_PROVIDER in .env

2. Soft Skills Extraction

• Hybrid approach: heuristic + LLM enhancement
• Semantic understanding of implicit skills
• Confidence scoring with evidence trails
• Maps to company value framework

3. Comprehensive Red Flags

• Job hopping with context analysis
• Experience gaps detection
• CV quality scoring
• Skill progression tracking
• Overqualification/underqualification assessment
• Context-aware rather than binary

4. Hybrid Feedback Generation

• Template-based: Fast, structured feedback
• LLM-enhanced: Personalized insights and improvements
• Interview prep: Suggested talking points for recruiters
• Growth path: Actionable next steps for candidates

5. Beautiful UI

• Streamlit-powered responsive interface
• Tinder-like candidate cards with match %
• Color-coded matching (🟢 🟡 🔴)
• Expandable detail feedback views
• Company profile customization

Usage

Step 1: Set Company Profile

1. Navigate to "🏢 Company Setup" tab
2. Enter mission, values, skills needed, role description
3. Click "Save Company Profile"

Step 2: Upload CVs

1. Go to "📤 Upload CVs" tab
2. Either upload JSON CV files OR load sample CVs
3. Click "Load and Process CVs"

Step 3: View Results

1. Navigate to "🎯 Results" tab
2. See candidates ranked by match %
3. Click "View Details" for full feedback breakdown

API Structure

Company Profile Format

{
  "name": "Company Name",
  "mission": "...",
  "values": ["Innovation", "Ownership", ...],
  "focus_skills": ["Leadership", "Communication", ...],
  "role_description": "..."
}

CV Format

{
  "name": "Candidate Name",
  "degree": "MS Computer Science",
  "university": "MIT",
  "years_experience": 7,
  "experience": [
    {
      "company": "...",
      "role": "...",
      "duration": "2 years",
      "description": "..."
    }
  ],
  "skills": ["Python", "Leadership", ...]
}

Match Score Components

• Technical: 35% - Technical skills match
• Soft Skills: 30% - Soft skills alignment
• Culture Fit: 20% - Values alignment
• CV Quality: 15% - CV structure & metrics

Performance

• Batch processing: Up to 20 CVs at once
• LLM abstraction: Supports any provider
• Feedback generation: ~5-10 seconds per CV
• UI response: Real-time feedback with Streamlit

Next Steps (Post-Hackathon)

• ATS integration (Workday, Lever, Greenhouse)
• Coaching recommendations
• Skill marketplace connection
• Analytics dashboard
• Resume optimization suggestions
• Salary benchmarking
• Retention prediction

Technologies Used

• Backend: Python, Pydantic
• LLM: Mistral API (with abstraction for others)
• Frontend: Streamlit
• Data: Pandas, JSON
• Deployment: Cloud-ready architecture

Submission Details

Project: Inter-Sight
Hackathon: Hack-Nation Global AI Hackathon
Date: November 8-9, 2025
Challenge: SAP SkillSense - AI for Talent Identification

Authors

Inter-Sight Team - Hack-Nation Hackathon 2025

License

MIT License - See LICENSE file for details

---

Support

For issues or questions:

1. Check documentation in /docs folder
2. Review code comments
3. Check .env.example for configuration help

Built with ❤️ for Hack-Nation 2025