• I am exploring the applications of active semantic segmentation strategies for mapping geohazards from SAR imagery to reduce the annotation effort.• I am investigating vulnerability to human life from debris flow hazards on a global scale.

This doctoral research aims to explore supervised deep-learning strategies for mapping landslides by exploiting Earth observation (EO) images sourced from different sensors – LiDAR, optical, and Synthetic aperture radar (SAR). Our multi-modal UNet-based segmentation models enabled us to set up automated landslide mapping pipelines in local and cloud computing systems that can generate regional scale inventories (≥1000 km sq), typically in the time frame of a few hours.This doctoral research was a part of the EC Horizon2020 BETTER project. We contributed to solving the geohazard challenge and deployed an automated mapping pipeline at TerraDue’s Ellip cloud platform.

EO Satellite Mission — RESOURCESAT 2A (launched on 7 December 2016)• Worked in the radiometric correction module of RESOURCESAT-2A's LISS-III, LISS-IV, and AWIFS-A/B imaging sensors. Developed statistical methods for correcting the imbalance and non-linearity in sensor response using data from pre-launch Light Transfer Characteristics (LTC) experiments. • Developed strategies/algorithms to fill data gaps observed in images after the launch. These gaps were caused by the malfunctioning of a few detectors in the pushbroom arrays.Planetary Mission — Mars Orbiter Mission (MOM)• Contributed to the development of a using data from Mars Orbiter Mission.

Consultancy Project — Study of Mine‐Wise Surface Runoff and Preparation of Management Plan for 15 Opencast Mines of Mahanadi Coalfields Limited.• Solved the problem of local mining companies by generating monsoon runoff models to estimate the volume of rainwater overflowing into and away from the active mining area. In the process, we identified an untapped potential of water storage in the abandoned mine workings, which can also act as a source for groundwater recharge.• The outcome of runoff modeling was appreciated by the State Pollution Control Board, which further recommended other mining companies to conduct similar studies.Teaching (Theory)• Remote Sensing and its Application• Geographic Information System

• For my Master's thesis, I used data products from Hyperion (hyperspectral) and MODIS/ASTER Airborne Simulator (only thermal channels) to generate lithological maps for two separate study areas. • Teaching Assistantship: I assisted Prof. Alok Porwal in conducting his regular theory courses and grading examinations.