Dharani Deivasihamani

Experience

• 

  Incture Technologies

  Jul 2017 - Aug 2019

  Associate Consultant

  - Developed User Interfaces for web applications using SAPUI5 framework (JavaScript, XML, CSS)- Lead the team in charge of UI deliverables for a client project

• 

  Technical University of Munich

  Apr 2021 - Dec 2021

  - Analyzed the feasibility of the usage of existing EUMETSAT atmospheric cloud mask algorithms on ESA AllSky7 camera network (Europe) images of the sky.- Co-authored the conference publication "Preparing the Characterization of the AllSky7 camera system" for the Proceedings of the International Meteor Conference, October 2022, contributing to the analysis of weather conditions and clouds for the best functionality of the system. - Developed Cloud Masks using CNN models that significantly detect and mask clouds from satellite imagery by fusing both SAR data from Sentinel 1 and NIR, SWIR, RGB and Coastal Aerosol data from Sentinel 2 satellites.- Evaluated the data analysis pipeline and visualized preliminary results for the publication: “Explaining the Effects of Clouds on Remote Sensing Scene Classification” in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2022 Tutored over 30 students of successive batch in downloading and usingSentinel constellation data from Copernicus Sentinel Hub using APIs to beused for creating remote sensing prototypes.

  • Graduate Student Intern - Institute of Astronautics

    Oct 2021 - Dec 2021

  • Student Assistant - Institute for Data Science in Earth Observation

    May 2021 - Sept 2021

  • Student Assistant - Institute of Astronomy & Physical Geodesy

    Apr 2021 - Aug 2021
• 

  German Aerospace Center (DLR)

  Jan 2022 - Jul 2022

  Master Thesis

  - Developed Convolutional Neural Networks (CNN) based object detection models for the identification of damaged buildings from multispectral satellite and aerial imagery used for emergency mapping during naturalcalamities.- Created a time-sensitive deep learning model with comparable results to SOTA, is robust and generalizable across large geospatial datasets of different spatial resolutions, ground sampling distances and acquisition geometry.- Published in DLR e-library, 2022 as "Performance and Transferability Assessment of Convolutional Neural Network (CNN) Based Building Detection Models for Emergency Response" Show less

• 

  Hochschule München University of Applied Sciences

  Feb 2023 - Mar 2024

  Research Assistant

  - Co-authored a conference publication titled "Using Node-RED as a Low-Code Approach to ModelInteraction Logic of Machine-Learning-Supported eHMIs for Virtual Driving Simulator Carla,"presented at the AutomotiveUI conference in 2023.- Supervised a Bachelor Thesis on “Traffic Light Detection & State Estimation from Drone Imagery”- Research on different object detection methods on RGB data from Aerial imagery

• 

  Azure Knowledge Corporation

  Jun 2024 - now

  Geospatial Machine Learning Expert