- End-to-end image captioning model using encoder-decoder architecture.
- ResNet-50 CNN for feature extraction + LSTM for caption generation.
- Trained on MS COCO dataset (82K+ images).
Python • PyTorch • ResNet-50 • LSTM • OpenCV
- Built CNN pipeline entirely in Python without ML libraries.
- Implemented convolution, activation, loss, and backpropagation.
- Used SGD optimizer for training.
Python • Custom Math • SGD
COLLEGE PLACEMENT PREDICTOR
- Predicts student placement outcomes.
- Feature engineering + Logistic Regression.
Python • Pandas • NumPy • Scikit-learn
- Exploratory analysis of student placement dataset
- Examined CGPA, internships, projects, and skills
- Used statistical methods and visualization
Python • Pandas • NumPy • Scikit-learn
RETRO DEVELOPER PORTFOLIO
- Designed and developed a personal developer portfolio with a retro pixel-style interface.
HTML • CSS • Responsive Design • Pixel UI Styling