title	TalentLens AI
emoji	🎯
colorFrom	blue
colorTo	indigo
sdk	streamlit
sdk_version	1.32.0
python_version	3.11.4
app_file	app.py
pinned	false

🎯 TalentLens AI: Intelligent Candidate Discovery Engine

Welcome to the TalentLens AI repository! This project is our official submission for the Redrob Intelligent Candidate Discovery & Ranking Hackathon.

We have built a deterministic, blazing-fast semantic ranking engine that processes candidate profiles, eliminates AI-generated honeypots through strict logical validation, and ranks candidates based on their contextual fit for the Senior AI Engineer role.

---

✨ Key Features

• 🛡️ Ironclad Honeypot Detection: Automatically flags synthetic trap candidates using temporal contradictions (e.g., active before signup) and logical impossibilities (zero-duration expert skills).
• 🧠 Semantic Skill Matching: Evaluates core backend technologies (Python, Java, Go) alongside modern AI competencies (LLM, RAG, Prompt Engineering).
• ⚡ High-Performance Pipeline: Processes a 100,000-candidate JSONL dataset deterministically under the 5-minute CPU constraint without requiring any external LLM APIs or GPU acceleration.
• 🎨 Premium UI Dashboard: Features a glassmorphism-inspired Streamlit web app with dynamic metrics and visual data validation.

---

🏗️ Architecture Overview

Our scoring algorithm evaluates valid candidates across four rigorous pillars to output a normalized score (0.000 to 1.000):

1. Skill Match (40%): Jaccard-style similarity weighted heavily towards advanced/expert AI proficiencies.
2. Experience Match (30%): A Gaussian distribution scoring model perfectly tailored for the 5-9 Years of Experience sweet spot.
3. Role Relevance (10%): Contextual bonuses for relevant current/past titles (e.g., "Machine Learning Engineer").
4. Platform Signals (20%): A composite score evaluating profile completeness, GitHub activity, interview attendance, and recruiter responsiveness.

---

🚀 Setup & Installation

This project is built using standard Python and requires no external API keys.

1. Clone the repository:

   git clone https://github.com/anishedu2234-coder/TalentLens-AI.git
   cd TalentLens-AI

2. Install the dependencies:

   pip install -r requirements.txt

---

💻 Reproducing the Submission

To generate the exact submission.csv from the 100,000 candidate pool used for our final hackathon evaluation:

python main.py

Note: The candidates.csv (JSONL format) and job_description.txt must be located in the root directory. The pipeline is heavily optimized for CPU execution.

---

🌐 Live Sandbox Demo

We have deployed a live, interactive version of the ranking engine for evaluation purposes.

👉 Launch the TalentLens AI Dashboard

• Upload a candidate JSONL dataset (up to 500 MB).
• View real-time analytics on honeypot detection.
• Export the strict, hackathon-compliant submission.csv directly from the web interface.

---

Built with ❤️ by Team STRIVERS