🤖 SkillSync AI: Next-Gen Recruitment Intelligence System

SkillSync AI is a professional-grade recruitment assistant that leverages Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) to automate resume screening, technical interview preparation, and ATS optimization.
Designed with a modern "Indigo & Slate" minimal interface, this tool provides data-driven insights for high-end IT recruitment environments.

🔗 Live Demo

Try the app here: SkillSync AI on Hugging Face Spaces

🚀 Key Features

• Multi-Model Intelligence: Model-agnostic architecture supporting OpenAI (GPT-4o), Google Gemini (1.5 Pro), Anthropic (Claude 3.5), and Groq (Llama 3.1).
• Semantic Tech-Audit: Scores resumes (0-100) against industry-standard IT roles or custom Job Descriptions (JDs) using semantic vector similarity.
• Multi-Candidate Leaderboard: Batch processing capability to upload up to 5 resumes simultaneously, ranking them from best-to-worst based on specific role alignment.
• Contextual RAG Engine: Ask specific questions about a candidate's history, such as "Does the applicant have production experience with AWS?"
• AI Interview Architect: Generates personalized Technical, Scenario-based, and Behavioral questions based specifically on the candidate’s unique background.
• ATS Optimization Engine: Identifies technical gaps and provides "Before & After" examples to enhance resume formatting for Applicant Tracking Systems.
• Enterprise UI: A minimal, professional dashboard designed for high-end IT recruitment environments.

🛠️ Tech Stack

• Frontend: Streamlit (Custom Indigo/Slate Theme)
• Orchestration: LangChain 0.3 (RetrievalQA, Multi-Model Router, Text Splitters)
• LLM: OpenAI GPT-4o, Google Gemini 1.5 Pro, Anthropic Claude 3.5 Sonnet, Groq (Llama 3.1 / Mixtral).
• Embeddings: HuggingFace all-MiniLM-L6-v2 (Local Processing)
• Vector Store: FAISS (Facebook AI Similarity Search)
• PDF Processing: PyPDF2
• Deployment: Docker & Hugging Face Spaces

🏗️ Architecture Workflow

1. Ingestion: Extracts raw text from PDF/TXT resumes using specialized parsers.
2. Vectorization: Splits content into semantic chunks converted into high-dimensional vectors via local embeddings.
3. Storage: Chunks are indexed in a local FAISS vector store for sub-second retrieval.
4. Intelligence Layer:
   • Multi-Model Router: Dynamically switches between AI providers based on user preference.
   • Audit Mode: Performs semantic scoring across specific technical competencies.
   • Ranking Mode: Iterative semantic evaluation of multiple documents to produce a comparative leaderboard.
   • Q&A Mode: Uses a RetrievalQA chain to provide grounded answers based only on the document.

🚀 Deployment (Hugging Face Spaces)

This project is containerized with Docker and deployed on Hugging Face Spaces for high-performance AI hosting.

⚙️ Installation & Setup

1. Prerequisites

• Python 3.11+
• An API Key from OpenAI, Google AI Studio, Anthropic, or Groq.

2. Clone the Repository

git clone https://github.com/Shiwam-m/SkillSync-AI
cd skillsync-ai

3. Setup Virtual Environment

python -m venv venv
# Windows:
venv\Scripts\activate
# Linux/Mac:
source venv/bin/activate

4. Install Dependencies

pip install -r requirements.txt

5. Run the Application

streamlit run app.py

📸 Usage Guide

• Configure: Select your preferred AI Provider and enter your API key in the sidebar.
• Upload: Drop a PDF resume into the "Resume Analysis" tab.
• Target: Select a predefined technical role (e.g., AI Engineer, DevOps) or upload a custom JD.
• Audit: Click "Analyze Resume" to generate the competency report.
• Batch Rank: Use the "Batch Ranking" tab to compare up to 5 candidates at once for a specific role to find the best fit quickly.
• Optimize: Use the "Interview Questions" and "Resume Improvements" tabs to prepare for the hiring process.

📜 License

• This project is licensed under the MIT License.

[IMPORTANT]

• Disclaimer : SkillSync AI is an assistant tool. Automated results should always be validated by human subject matter experts.