Automated Catalog of Long Period Variables with the Palomar Gattini-IR

under Prof. Mansi Kasliwal, Caltech

I developed a gradient boosted decision tree classifier to sift out Long Period Variables (LPVs) from heavily class-imbalanced lightcurves from Palomar Gattini-IR achieving 96% accuracy and a g-mean score of 0.969. I also assisted in observing interesting candidates flagged by the classifier with the Triple Spectrograph instrument on Palomar Observatory.

Real-Bogus Astrophysical Transient Classifier for GROWTH-India Telescope

under Prof. Varun Bhalerao, IIT Bombay

I built a deep learning model to classify the transient events obtained through image subtraction on GROWTH-India Telescope data. The science, reference and difference images are used as inputs to the neural network, which gives a real-bogus score between zero (for bogus) and one (for real) as output. The model implementation is done using the Keras API of TensorFlow.

Gamma Ray Burst Injection in CZTI Datastream

under Prof. Varun Bhalerao, IIT Bombay

I developed a pipeline to inject artificial Gamma Ray Burst signals into lower level event data of the Cadmium Telluride Zinc Imager (CZTI) aboard AstroSat, to quantify the efficiency of data analysis pipeline using T90 calculations. I also automated untriggered searches for GRBs using the CZTI Interface for Fast Transients (CIFT).

Automated Identification of Solar Flares

under Indian Space Research Organisation (ISRO)

As part of a team of 6, developed SuryaDrishti, a standalone web-based application using Python and Angular to identify and categorise X-ray bursts from Solar X-ray Monitor aboard Chandrayaan-2 based on peak energy flux, with functionalities to aid flare analysis.