Kaustubh Javkar
Climate-Driven Wing Morphology in Erebia
Butterflies
MS thesis · May 2025 – May 2026
Studied how climate shapes wing morphology in Erebia
butterflies, using deep-learning-based shape analysis
(SAM2 feature embeddings) combined with species
distribution modelling. Found that forewings diverge
sharply between species while hindwings stay more
conserved, driven mainly by precipitation seasonality,
temperature extremes, and elevation.
Deepjit Debnath
Network Topology of Indian Summer Monsoon Onset
Variability
MS thesis · May 2025 – May 2026
Built a network-based framework, combining event
synchronization, network divergence, propagation
analysis, and community detection, to compare the
climate structure of early versus late monsoon onset
years using 1979–2020 rainfall data. Found that the
two onset types produce distinctly different network
topologies, likely shaped by different phases of the
Boreal Summer Intraseasonal Oscillation.
Vishnu Kadam
Reproducibility and Regional Evaluation of SFNO-Based
Weather Forecasting
MS thesis · May 2025 – May 2026
Tested the reproducibility and regional performance of
SFNO-based ForeCastNet, a neural-operator weather model,
on both global and Indian-subcontinent data at two
resolutions. Built a full training-to-evaluation
pipeline and assessed forecast accuracy and stability
using RMSE and anomaly correlation across lead times.