AI-Powered HR Recruitment Platform: Practice of Integrating Machine Learning and Large Language Models

Core Project Overview

This project is the AI-powered-HR-Platform released by eyamkaouar on GitHub (June 7, 2026), an end-to-end AI recruitment platform. It integrates machine learning (salary prediction) and large language models (candidate matching/growth analysis) to solve recruitment pain points: inefficient screening of massive resumes, lack of objective standards for matching, and no data support for salary negotiations. The platform includes three core modules: intelligent resume parsing, salary prediction, and AI matching evaluation, along with a visual interface, providing HR with practical intelligent tools.

Industry Challenges and Solutions

Current recruitment faces three major problems:

1. Manual screening of massive resumes is time-consuming and labor-intensive, prone to omissions and misjudgments;
2. Candidate matching relies on keywords, lacking comprehensive and objective evaluation;
3. Salary negotiations lack data support, making it difficult to balance costs and expectations. This project is a fully functional end-to-end application that provides a systematic technical solution by integrating ML with traditional processes.

Three Core Modules

1. Intelligent Resume Parsing: Extracts over 120 features (skills, experience, education, etc.) based on pdfplumber, providing structured data to overcome the limitations of keyword matching;
2. Salary Prediction Model: Extra Trees regression model (reduces overfitting), predicts salary ranges by combining experience, skills, and region, providing HR with negotiation references;
3. LLM Matching Evaluation: Integrates Mistral AI to generate humanized reports, analyzing matching degree, career trajectory, and potential, explaining "why it's suitable" and solving the ML black box problem.

Tech Stack and Deployment

• Backend: FastAPI (asynchronous, automatic API documentation, suitable for ML services);
• Frontend: React+Vite, Tailwind CSS (responsive), Framer Motion (interactive effects), Plotly (data visualization such as global salary map);
• Deployment: Docker containerization, one-click startup with docker-compose, lowering the trial threshold.

Data Protection Design

1. Data Storage: Resumes are processed only in memory and not permanently stored, reducing leakage risks;
2. Report Presentation: Filters internal ML features and scores to ensure reports are professional and concise, avoiding technical interference with decision-making. Complies with modern data protection regulations.

Scenarios and Value

• Campus Recruitment Screening: Quickly identify high-potential fresh graduates;
• Executive Search: Data-driven salary suggestions to reduce talent loss;
• Internal Talent Inventory: Analyze team skills and growth to support training/promotion decisions. Core value: AI acts as an assistant to improve efficiency while retaining human final decision-making authority.

Insights and Outlook

• Insights: Modular design supports independent iteration (e.g., expanding resume formats, industry-specific salary models);
• Future: Multimodal video interviews, ATS integration automation, feedback loop to optimize algorithms;
• Conclusion: The project combines the rigor of ML with the flexibility of LLM to solve real pain points, serving as a practical reference for AI implementation in the HR field.