About me

Hi there! I am a PhD student and Scienctific Assistant at ETH Zürich in Photogrammetry and Remote Sensing department(PRS). I joined the EcoVision Lab in 2018. I completed my Masters in Informatics from Technische Universität München in 2018. After doing my Master Thesis at EcoVision Lab on the topic Flood estimation through semantic image interpretation in 2017-2018, I am continuing my work on the project. The main goal of the project is to use images collected from social media of various flood events and use them to quantify flood. For quantification of floods, different objects of known dimensions which are partially submerged in flood water are used.

BackGround

I completed my Bachelors from Delhi Technological University (New Delhi) in 2013. Following that, I worked for two years in Samsung Research Institute as Software Development Engineer. I worked on Android kernel platform and was part of System Memory and Tools Android File System Team. After that I joined TU München for my Masters in Informatics. I completed my Masters in September 2018 and following that I joined in November 2018, ETH Zürich as PhD student and research assistant.

Interests

During my Master’s studies, I developed an interest in the field of Computer Vision and Machine learning. Due to drastic climate change, there is a need for research in natural disasters and global challenges like Floods, Poverty, Tsunamis. I believe that academic research can have a remarkable influence on such problems. At EcoVision Lab, I work on one such project about Floods. During the first two years of my PhD we worked on how to quantify flood height using images collected from social media of various flood events.

Presently we are working on the research question, how can deep learning and physically-based hydraulic models be combined so that it includes physical constraints representations? Instead of learning all evidence from scratch, we aim at tightly integrating deep learning and physically-based hydraulic modelling to achieve the best of both worlds: physical interpretability, outputs with well-calibrated uncertainty estimates, and the (fast) predictive power of neural networks.

I also have considerable interest in natural language understanding topics like sentiment analysis, language modelling etc and have taken Data Mining Lab during my Masters and also Natural Language Understanding lecture and participated in the project for more experience and understanding of the subject.

Apart from the above, I enjoy travelling, reading books and have recently found a fondness for cooking.