Specializing in deep learning based modelling for noisy time‑dependent data and decision making. Advancing research in randomized neural architectures, ensemble methods, and uncertainty‑aware predictive modelling.
I am a PhD candidate at IIT Guwahati, with a background in applied mathematics and machine learning, focusing on neural network-based modeling of high-frequency, noisy time-series data for forecasting and decision-making.
Implementation of research ideas as reproducible experiments and scalable code.
Data‑driven approaches to modelling markets, risk, and trading behaviour.
Thesis title: Learning Frameworks for time series forecasting and financial decision-making using backpropagation and randomized neural networks.
Supervisors: Prof. Selvaraju Natarajan (Department of Mathematics, IIT Guwahati) and Prof. Ponnuthurai Nagaratnam Suganthan (Department of Computer Science and Engineering, Qatar University).
My doctoral research develops structured learning frameworks for time-series forecasting and sequential decision-making by integrating statistical modelling, randomized neural representations, contextual and memory mechanisms, state-space modelling, and reinforcement learning. The work focuses on volatility and risk forecasting, significant wave-height and load forecasting, portfolio optimization, and multi-stock trading under nonlinear, heteroscedastic, and regime-varying dynamics.
Master's thesis: Applications of Deep Learning in Supervised Data Analysis
Advisors: Prof. Manoj Thakur and Prof. Samar Agnihotri
Utilized Convolutional based GANs for image generation; deep networks (FNN, CNN, ResNets) for image classification; and RNN, LSTM, GRU, and Bi-LSTM models for financial forecasting.
Foundational training in mathematics and quantitative sciences.
Award: DST-Inspire Scholar (2016–2021), scholarship for meritorious students in Natural and Basic Sciences.
I develop learning frameworks for complex time-dependent data, with emphasis on uncertainty, efficient neural architectures, survival modelling, and decision-making under dynamic environments.
Models for nonlinear, heteroscedastic, and regime-varying sequential data, including volatility, wave-height, and load forecasting.
Randomized neural networks, RVFL-type architectures, deep ensembles, contextual mechanisms, and memory-enhanced learning.
Learning-based frameworks for portfolio optimization, stock trading, option pricing, hedging, and risk-aware financial decision-making.
Neural and randomized learning approaches for time-to-event modelling, survival prediction, and uncertainty-aware medical decision support.
Lead Teaching Assistant: DA105 (Linear Algebra); DA210 (Time Series Analysis & Forecasting).
Teaching Assistant: MA101 (Real Analysis and Calculus); MA683 (Statistical Analysis); MA373 (Monte Carlo Simulation); MA374 (Financial Engineering Lab); MA225 (Probability Theory and Random Processes); DA105 (Linear Algebra).
For research collaborations, seminar invitations, reviewing, or academic discussions on time series analysis and financial decision making using deep learning and reinforcement learning techniques, feel free to reach out using the channels below.
Department of Mathematics
IIT Guwahati – 781039
Assam, India