Built using PyTorch with CNNs and LSTMs. Achieved 93.78% accuracy on mixed datasets including FaceForensics++.
ML/DL system using BiLSTM model achieving 88% accuracy for duplicate question pair detection.
SaaS banking platform with Plaid & Dwolla integration for bank linking and funds transfer.
Real-time AI-generated voice detection using CatBoost achieving 93% accuracy with 0.002s inference.
