- Azure Files team under Azure Storage

•  Analysed and summarised the primary works of key authors to find the problem statement for Android Malware Detection System under the guidance of Professor Arash Habibi Lashkari•  Deduced a novel framework generating realistic adversarial examples incorporating Provably Certified Defense against Adversarial Attack•  Building a non-uniform manipulation set for android malware applications to design examples by analysing feature importance•  Applying Projected Gradient Descent (PGD) to optimise worst-case loss on the convex-outer bound to a non-convex set of final-layer activations of deep ReLu networks

• Worked under Azure Files team of Azure Storage• Proposed an optimised high-level design for the Force Close Handle Live Site Tool to mitigate leaked file handles of NFSV4 Azure File Shares • Implemented three commands in C++ among Partition Layer, NFS Frontend, Storage Diagnostics Server in Azure Storage architecture• Tested in local environment on handles being leaked, opened to NFS file share mounted by ubuntu client locally, by Storage Diagnostics client close command

• Conducted the virtualised Cyclothon, Chess, Quizzicals, Fitness challenges and Sports Analytics worshops online for more than 500 boarders• Increased the participation from all the batches in WIN India 2.0 sporting extravaganza and the hall secured bronze in virtual marathon

• Built a responsive Login,Signup and other related web pages for the employment web application using React JS and Reactstrap• Designed features like login from an external account, username authentication, logout, keeping the user signed in via API integration

• Part of the Mars Colonisation Program with Ace Hacker ,in Microsoft Engage 2020• Worked in a team of three members to build a web application deploying invincible Tic-Tac-Toe game using the Minimax algorithm up till 6th depth, optimised by Alpha-Beta pruning• Extended the order of matrices to 4x4 and 5x5 in addition to the 3x3 matrix with 4 difficulty levels, achieving scalability and generalizabilty• Added the feature of providing AI-powered suggestions for the next move to the two human players competing against each other• Designed an interactive and attractive frontend using Javascript, HTML and CSS and hosted the web application on Github pages