Resume/CV

MS Student Data Science Intern

Indian Institute of Tropical Meteorology, Pune, India

Developed spatiotemporal models to forecast environmental risk events (e.g., forest fires, stubble burning) using satellite data. Built ConvLSTM models achieving 0.8 temporal correlation for 1-day forecasting across geospatial grids. Designed and implemented automated data pipelines for scalable preprocessing and visualization. Enabled operational decision-making by translating real-world risk forecasting problems into ML solutions. Co-authored a peer-reviewed publication in Modeling Earth Systems and Environment.

• Fire Forecasting
• Deep Learning

MS in Engineering Science

University at Buffalo, Buffalo NY, USA

Data Science

• Introduction to Numerical Mathematics for Computing and Data Science
• Introduction to Probability Theory for Data Science
• Programming and Database Fundamentals for Data Scientists
• Statistical Learning and Data Mining I
• Statistical Learning and Data Mining II
• Data Models and Query Language
• Introduction to Machine Learning
• Data Intensive Computing
• Predictive Analytics

Post-Graduate Diploma in Applied Statistics

Indian Statistical Institute, Kolkata, India

Data Analytics

• Basic Statistics
• Basic Probability
• Statistical Methods
• Survey Sampling
• Introduction to Official Statistical System
• Statistics and Economy
• Introduction to R and Python
• Multiple Regression
• Advanced Regression
• Time Series Analysis and Forecasting
• Multivariate Statistics
• Statistical Machine Learning

Integrated Bachelor's and Master's of Science

Indian Institute of Science Education and Research, Pune, India

Science

Harnessing deep learning for forecasting fire-burning locations and unveiling PM2.5 emissions

Modeling Earth Systems and Environment

The paper applies spatiotemporal deep-learning architectures—specifically ConvLSTM and ConvGRU—to predict where crop-residue and forest fires will occur across India up to three days in advance. By coupling these forecasts with meteorological and vegetation indices, the authors produce dynamic PM2.5 emission estimates, achieving a correlation of ≈ 0.8 for next-day predictions and moderate skill for days two and three, suggesting a practical route to more accurate short-term air-quality forecasts.