Implemented SOTA ML models (Temporal Fusion Transformer, LSTM) to forecast thermospheric density.- Achieved a 10-15% reduction in prediction errors compared to current empirical models.- Developed a real-time continual learning framework that: - Automatically ingests new data, - Updates models, and - Benchmarks against previous versions and empirical modelsEstablished a robust and adaptive system to reduce model uncertainty and improves predictive accuracy.

Current ProjectsSolar Flare Identification- Developing a deep learning CNN model in Python to identify large solar flare events (M/X class) using solar images.- Integrated geospatial data from GOES and NASA's SDO to create a training dataset.- Created Python libraries to process GOES reports and extract solar images from a large SDO dataset on AWS, using tools like Zarr/Dask for efficiency.Automated Detection of Amorphous Bright Spots- Working on… Show more Current ProjectsSolar Flare Identification- Developing a deep learning CNN model in Python to identify large solar flare events (M/X class) using solar images.- Integrated geospatial data from GOES and NASA's SDO to create a training dataset.- Created Python libraries to process GOES reports and extract solar images from a large SDO dataset on AWS, using tools like Zarr/Dask for efficiency.Automated Detection of Amorphous Bright Spots- Working on identifying amorphous bright spots in UV emission images from Mars, using data from the - EMUS instrument on the EMM spacecraft.- Prepared composite Martian disk images from varied sources for analysis.- Employing multiple detection methods, including DBSCAN clustering and blob detection. Characterizing Gravity Waves in Mars Atmosphere- Utilizing a linear two-fluid model to study gravity wave fluctuations observed by MAVEN NGIMS.- Translated and optimized legacy IDL code to Python, reducing execution time significantly through vectorized computation.- Expanded analysis period significantly, employing optimized code and dynamic analysis functions.Responsibilities- Develop and refine models for atmospheric and geospatial phenomena using advanced machine learning.- Manage and integrate large datasets from varied sources, ensuring high data quality for analysis.- Apply data analysis and image processing techniques to uncover insights from complex datasets.- Lead research projects and writing proposals to secure funding.- Create tools and libraries to streamline data processing, focusing on efficiency and scalability.- Collaborate with interdisciplinary teams, providing data-driven insights for strategic decision-making.- Participate in educational outreach and mentorship, sharing knowledge and fostering scientific curiosity. Show less