Academic Projects
Neural Network-Based Solvers for PDEs in Biological Systems (Major Project)
- Research Aim: To explore physics informed neural networks (PINN) for solving PDE based biological models
- Methodology: Implemented a FNOs (Fourier Neural Operator) based PINN model, implemented using DeepXDE framwork, with tensorflow backend.
- Result: Modeled Fisher-KPP, Continumm tumor growth model, Reaction-diffusion 2d turing pattern, and Lotka-Voltera system using PINN and compared results with finite method solvers.

draft paper
- Research Aim: To build AI/ML based model for predicting fungal SM bioactivity.
- Methodology: Implemented an ensemble of machine learning model based on processed BGC (Biosynthetic Gene Clusters) for binary classification of SM activity.
- Result: Achieved accuracy of 83 % for SM classification task on the fungal BCG data.
ML driven analysis of genetic factors contributing to RIF in IVF (NTCC Project)
- Research Aim: To build Machine Learning model based on metagenomic data and deriving insights on genetic factors contributions.
- Methodology: Processed 16s RNA metagenomic data using bioinformatics pipeline, into a numeric feature matrix for predictive modeling. Implemented ensemble of ML models for robust prediction of RIF status.
- Result: Achieved accuracy of 89% on classification of RIF patients. Ranked the genetic factors contributions by random forest model’s feature importance metric.
