Web dog CNN(92% accuracy )

Created a CNN-based web application to classify dog breeds with an accuracy of 92%

GAN

Developed a custom neural network library, featuring a dynamic computing graph. Utilized this library to successfully implement a Generative Adversarial Network model

NeedleDream(Online course)

Developed site for paid online course with 300+ users Conducted sentiment analysis on reviews and created analytical models from raw data Increased sales and customer satisfaction by 15% Improved metrics (CV1 +5%, CPC +8%) through clustering and social media analysis Increased completion rates by 15% and user satisfaction by 20% through engagement metric analysis