6 projects - click any card to expand details.

Computer Vision

Brain Tumor Detection

A CNN-based deep learning model that detects brain tumors from MRI scans using TensorFlow and Keras. Achieves high accuracy through careful preprocessing and architecture design.

CNNDeep LearningTensorFlowMedical Imaging

Teaching angle

A live demonstration of how CNNs learn spatial features from medical images - ideal for teaching convolutional architectures and real-world AI application.

Key concepts

Convolutional LayersTransfer LearningBinary Classification
Regression Analysis

House Price Prediction

Compared over 8 regression models including Linear Regression, Ridge, Lasso, Decision Trees, and Gradient Boosting to predict house prices with Python and scikit-learn.

RegressionScikit-learnFeature EngineeringModel Comparison

Teaching angle

A comprehensive comparison of 8+ regression algorithms - perfect as a classroom case study for understanding model evaluation, bias-variance tradeoff, and feature selection.

Key concepts

Bias-Variance TradeoffRegularizationCross-Validation
Backend API

Student Management System

A production-ready RESTful API built with FastAPI and SQLAlchemy for managing students, courses, and enrollments. Features JWT authentication and full CRUD operations.

FastAPISQLAlchemyJWTREST API

Teaching angle

A working demonstration of software architecture for students learning API design, database normalization, and authentication - real-world software engineering in practice.

Key concepts

REST ArchitectureDatabase DesignAPI Security
Healthcare AI

Heart Disease & Diabetes Classification

Built classification models for heart disease and diabetes prediction using logistic regression, SVM, and ensemble methods with comprehensive evaluation metrics.

ClassificationHealthcareSVMEnsemble Methods

Teaching angle

Using real medical datasets to teach classification algorithms, precision vs recall tradeoffs, and the ethical responsibility of AI in healthcare.

Key concepts

Classification MetricsPrecision/RecallData Imbalance
Anomaly Detection

Credit Card Fraud Detection

Developed an anomaly detection pipeline for identifying fraudulent transactions from highly imbalanced financial data using resampling and ensemble classifiers.

Anomaly DetectionSMOTEImbalanced DataFinance

Teaching angle

Teaches students about imbalanced datasets, SMOTE, and anomaly detection - critical challenges in real-world machine learning that textbooks often skip.

Key concepts

Class ImbalanceSMOTE OversamplingROC-AUC Curve
Computer Vision

MNIST Digit Recognition

Implemented digit classification on the MNIST dataset using fully connected and convolutional neural networks, demonstrating the progression from simple to deep architectures.

Neural NetworksMNISTTensorFlowComputer Vision

Teaching angle

The classic "hello world" of deep learning - used to teach neural network fundamentals, backpropagation, and how machines learn to see.

Key concepts

BackpropagationActivation FunctionsSoftmax Classification