Predicting student performance from lifestyle habits and delivering actionable career guidance remain critical yet largely disjoint challenges in engineering education. Existing approaches typically address these problems independently and often rely solely on academic indicators, lacking deployable, student-centric interfaces. This study presents a unified, end-to-end web platform that integrates performance prediction with intelligent career guidance. The system is trained on a 1,000-record Kaggle lifestyle dataset (ages 17-24) comprising 14 features, including sleep patterns, social media usage, dietary habits, and mental health indicators, along with an augmented dataset of 5,000 records. Multiple machine learning models were evaluated, including Logistic Regression, K-Nearest Neighbors (KNN), Random Forest, XGBoost for classification, and Multiple Linear Regression (MLR), Ridge, and Lasso for regression, across three train-test splits (70:30, 80:20, and 90:10). Results demonstrate that Random Forest and XGBoost achieved perfect classification performance with 100% accuracy and ROC-AUC of 1.000 across all datasets and splits. For regression, MLR achieved an R2 of 0.8955 with an RMSE of 5.41 on the augmented dataset, outperforming comparable approaches without requiring complex feature engineering. The platform further incorporates a nine-agent GPT-4o-mini-based career intelligence pipeline, enhanced with live job-market insights via the Adzuna API and deterministic fallback mechanisms. The proposed system uniquely combines real-time predictive analytics, adaptive career recommendations, and a fully deployed three-tier architecture, making it a comprehensive solution for modern student support systems.