Research Scholar, Siksha O Anusandhan (Deemed to be) University (Jan 2022 – Present)
Fine-tuningConducted AI research building hybrid deep learning models for financial time-series prediction using CNN, GRU, attention mechanisms, and transformer-based architectures. Developed multi-modal modeling approaches by integrating sentiment signals with market indicators, requiring preparation and alignment of labeled/derived data inputs for supervised learning. Performed feature engineering, temporal validation, and model optimization to improve predictive robustness for financial analytics tasks. • Designed hybrid CNN-attention-GRU and Transformer-based architectures for stock/financial sequence prediction. • Integrated sentiment signals with market indicators in multi-modal learning frameworks. • Implemented residual learning and attention-based feature weighting to enhance prediction robustness. • Conducted temporal validation and optimization for financial datasets using model training workflows.