- Developing coding skills in Python for high-performance scientific computing, geoscientific simulations, and data analysis.
- Learning to implement and integrate Geodynamic models (ASPECT, DOUAR, FANTOM), Landscape evolution models (FastScape, Landlab) and Thermochronologic models (PeCube, HeFty) for simulating tectonic-climatic feedbacks.
- Exploring physics-informed machine learning (PIML) methods for modeling complex Earth systems involving multiple physical processes, scales, and phases. PIML combines deep learning with fundamental physical laws, aiming to create physics-consistent and scalable models that facilitate scientific discoveries from data.
- Working on building geoscience-specific AI tools (Artificial Intelligence for Geoscience) that include - Geoscience Language Models (GLM) for extracting structured knowledge from domain-specific texts and datasets, and Graph Representation Learning (GRL) to integrate and analyze gravity, magnetic, geochemical, and remote sensing data for critical mineral exploration in deep-cover, data-sparse regions.
Mineral-Microbe-Water Interaction (2024-present)
- Exploring (a) how minerals react with surrounding solutions—focusing on atomic to pore-scale geochemical processes for understanding soil-carbon dynamic, (b) the biogeochemical transformation of minerals and transport of chemical components across different spatial and temporal scales to study environmental contamination, and (c) models of ecological and biogeochemical systems that assess the impacts of mineral-microbe-water interactions in geoengineering methods such as Enhanced Rock Weathering for CO₂ sequestration in soils.
Earth Science Education Research (2023-Present)
- Mentoring, Outreach & Experimental Training: At Prayoga, through the Anveshana Research program, I mentor high school students via project-based learning. This includes guiding field-based geoscience projects and microbiology lab experiments, promoting hands-on, interdisciplinary learning. These efforts aim to popularize Earth Sciences and cultivate the next generation of geoscientists.
Bayesian Probabilistic Modeling (2020-Present)
- Applied Bayesian Probabilistic Modeling (BPM) for inverting detrital cooling ages into erosion history using PyMC Python library.
- BPM is an approach in artificial intelligence that seamlessly integrates statistical inference with machine learning. Unlike traditional machine learning, which primarily emphasizes predictions, BML incorporates the principles of probability and inference, creating a framework where learning dynamically evolves as new evidence is acquired.
Tectonic-Geomorphology, Thermochronology (2014-Present)
- Studied the interactions between climate and tectonic processes associated with the Himalayan orogeny.
- Tools used in quantifying climate-tectonic interplay include (1) geochronology, such as low-temperature thermochronology and optically stimulated luminescence dating; (2) remote sensing data; (3) multidisciplinary sedimentology, tectonics, and geomorphology approach; and (4) Bayesian statistical modeling.
- Field expeditions include Arunachal Pradesh Himalaya (specifically in the Siang, Dibang, and Lohit river valleys), the Shillong plateau, and the Belt of Schuppen (especially in the southeastern boundary of Assam valley and the western part of Nagaland).
