My Projects

Transfer learning with pretrained VGG16 and ResNet architectures for medical image analysis, achieving 92% validation accuracy.

Automated neural architecture search techniques for efficient model design with multi-objective optimization, achieving 30% latency reduction.

Multi-region deep-learning system for optimal power generation forecasting using environmental, temporal, and demand data.

Real-time vision-based system for American Sign Language gesture recognition with optimized CNN architectures, achieving 95% classification accuracy.

Production-grade machine-learning–driven WAF for real-time threat detection, anomaly identification, and automated security response.

LSTM-based sequence modeling system for fake news detection using contextual understanding and pretrained GloVe embeddings, achieving 89% F1-Score.