Built a responsive frontend with Next.js featuring animated transitions and interactive dashboards. Integrated ApexCharts to visualize SEO metrics like keyword rankings and page performance. Implemented a fast login system focusing on performance and accessibility.

Developed a Next.js frontend with WebSocket integration for real-time updates and notifications. Built interactive analytics dashboards using Framer Motion for smooth transitions. Designed scalable UI with optimized rendering using dynamic routing and ISR.

Built an XGBoost model to predict SEO scores using Lighthouse metrics and features like keyword density, achieving R² score of 0.82. Processed web-crawled data with Pandas and Scikit-learn for feature engineering and model training.

Developed a CNN-based deep learning model for age and gender prediction from facial images. Achieved 88%+ accuracy for gender classification and MAE < 5 years for age prediction using TensorFlow and multi-task learning on UTKFace dataset.

Implemented the famous Transformer model for machine translation, achieving BLEU score of 28.5 on WMT 2014 English-German dataset. Built with PyTorch using multi-head self-attention, positional encodings, and Hugging Face tokenization.

Built a 1D CNN model for product title classification achieving 87%+ accuracy on Torob dataset. Designed efficient multi-layer architecture with embedding, convolutional, and pooling layers using PyTorch for fast training and inference.

Developed a ResNet50-based CNN model for product image categorization achieving 90%+ accuracy. Trained on diverse Torob dataset featuring fashion and product images with robust data augmentation and transfer learning techniques.

Built a VGG16-based CNN model for emotion classification from facial images using FER-2013 dataset. Implemented face detection with OpenCV and fine-tuned pretrained model with data augmentation for multi-class emotional state recognition.