-Estimation of Optimal Parameters for Probabilistic Models of Collision Centrality Distributions using Machine Learning Techniques.- Applied the above techniques on Pb-Pb Collision ATLAS dataset

Course Name: Introduction to Data Science II & III (for Engineering Undergraduates)Professor Name: Prof. A. D. DileepTenure: 1 year- Evaluated assignments for 72 students , under the guidance of Prof. A.D. Dileep- Lab Teaching Assistant for these students

- Explored control flow analysis on library codebase and obtain knowledge graph. - Used probabilistic models to learn entity relation embeddings for link prediction and related tasks. - Implemented latest architectures in the field of embedding learning in PyTorch. - Proposed a new framework of learnable non euclidean background geometry for modelling embeddings.

Used recurrent neural networks (LSTM) to generate probable tests for external service application which a user can execute and experimented with other shift-right approaches to software testing.

1. Introduced new derivations related to path integral formulation of collapse models to solve quantum measurement problem.2. Stated that the zero limit of quantum mechanics is not just hbar tends to zero but also some mechanism to kill the superposition for systems which have macroscopic dimensions.3. Used collapse models (GRW, QMUPL) and path integral approach to deduce that all particle dynamics description given by the above collapse models results in standard quantum mechanics and classical statistical mechanics under appropriate limits on the parameters (collapse rate and localisation width) of the model.