Developed an end-to-end AI-powered system for lung cancer detection and classification using histopathology biopsy images, leveraging deep learning and explainable AI techniques. 🧠 Built and compared three models (Custom CNN, VGG16, ResNet50) using transfer learning, achieving up to ~98% accuracy with ResNet50. 🔬 Designed the system to classify images into: • Lung Adenocarcinoma (Malignant) • Squamous Cell Carcinoma (Malignant) • Normal Lung Tissue (Benign) 🔥 Integrated Grad-CAM (Explainable AI) to visualize model attention, enabling interpretability of predictions for medical insights. 📊 Implemented complete evaluation pipeline: • Confusion Matrix, ROC Curves, Precision/Recall/F1 • Balanced dataset (15,000 images – LC25000) • Data augmentation & optimized tf.data pipeline 🖥️ Developed a production-ready Streamlit web app with: • Real-time image upload & multi-model predictions • Confidence scores & probability visualization • Grad-CAM heatmaps for diagnosis explanation • Batch analysis & downloadable reports ⚙️ Tech Stack: Python, TensorFlow/Keras, OpenCV, Streamlit, NumPy, Pandas, Matplotlib, Seaborn 💡 Impact: This project demonstrates how AI can assist medical professionals by improving diagnostic efficiency, accuracy, and interpretability in cancer detection. ⚠️ Note: Built for research & educational purposes (not for clinical use).