Employee Guard System

Project Name: EmployeeGuard - End-to-End AI System

EmployeeGuard is a comprehensive End-to-End Machine Learning System designed to help Human Resources (HR) departments predict employee attrition risk and analyze the key drivers behind employee turnover, enabling proactive decision-making and improving retention strategies.

Key Development Stages:

1. Data Science & Machine Learning Pipeline:

- Performed Exploratory Data Analysis (EDA) using Pandas and Seaborn to identify critical retention factors such as OverTime work.

- Addressed Class Imbalance and engineered automated Data Preprocessing pipelines using ColumnTransformer, StandardScaler, and OneHotEncoder.

- Evaluated multiple classification models and optimized a Balanced Logistic Regression using Hyperparameter Tuning (GridSearchCV) and Threshold Adjustment to maximize the recall score for identifying high-risk employees.

2. Backend API Development (FastAPI):

- Developed a high-performance RESTful API using FastAPI to seamlessly bridge the trained ML model with the client application.

- Exposed key endpoints (/predict, /analytics, /model-metrics) to deliver real-time predictions and dynamic analytics.

- Integrated Pydantic for data validation and request schema verification.

3. Interactive Frontend Dashboard:

- Designed a modern, responsive web application using HTML5, CSS3, and JavaScript.

- Built live data visualization cards and dynamic charts powered by Chart.js to present model performance metrics (Confusion Matrix, Feature Importance) and key HR indicators.