Publications:

Projects

Upcoming:

  • Testing for Machine Learning applications
  • Continuous Improvement
  • Data Version Control

Running:

  • Learning basic concepts of rust by designing a simple web service (Blog and code coming soon)

Completed:

  • Self Supervised Out of Distribution detection for Medical applications(Master thesis) {October 2019 - July 2020}: The main aim of the thesis is to use Out of distribution detection to detect anomalous brain CT scans. Self Supervised learning is employed for efficient utilization of unlabelled data.

  • Inverse problems in PDE-Driven process using deep learning (Research project) {April 2019 - October 2019}: Proposed a Neural solver for learning optimal iterative scheme of the Partial differential equation using data driven methods. The trained neural solver results in a faster convergence compared to normal Finite difference method and Finite element method solvers for the same PDE.

  • Stroke detection using BRATS dataset {May 2019}: Using an Autoencoder like structure to segment the strokes in BRATS dataset. The implementation is inspired by FCN32 but instead of the interpolation after feature extraction used a Convolution transpose to reconstruct the image.

  • Network analytical take on European parliament (Praktikum) {October 2018 - Feb 2019}: Objective is to find hidden agenda from the speeches of European parliament.My contribution is to build a topic model from the extracted speeches and then use this to create a graph model that finds a hidden coalition using Hidden community detection.

  • Physics aware GAN (Course project) {April- July 2018} Application of Generative Adversarial Network on Velocity simulation of a smoke flow.The goal was to extend the simulation on higher resolution while training on a lower resolution.

  • Autoencoder for velocity images {May 2018}: Simple autoencoder network to learn the distribution of Velocity smoke images.

  • Free surface simulation solver {April-July 2018}: A 2d navier stokes solver written in C which can handle aribitrary geometries, heat transport and free surface simulation. Experiments were conducted on a tanker with different conditions to check solver application.

  • CFD analysis of Co-Flow Jet Airfoil (Bachelor Project) {August 2015- May 2016}: A control flow technique applied on a non-symmetric airfoil. The objective to increase the lift coefficient and drag coefficient achieved successfully.

  • Case study of Low power Vapour compression Refrigeration (Research Project) {November 2014 - May 2015} A numerical simulation of various components of the refrigerator to obtain the temperatures of the refrigerator and to obtain the Coefficient of performance of the refrigerator for the experimental conditions obtained from the Vapor Compression Testing apparatus in the Refrigeration lab.