Develop astrophysics data processing algorithms, prototype, and production-quality software from real and simulated space mission data for NASA's Nancy Grace Roman Space Telescope for the following science operations:

• Identification of microlensing events to discover exoplanets.
• Characterization of exoplanet features from microlensing events.

Maintained and extended a Python-based ETL data pipeline for an ML/AI framework that ultimately estimates power plant GHG emissions based on satellite imagery, atmospheric chemistry datasets, and real-time grid operations.

• Extract: Built scripts and continuous integration to ingest time series data for power plant generations and emission sources and satellite imagery (Google Earth Engine and Planet) at a rate of 100 GB per day.
• Transform: Developed a data pipeline that transforms data in batch or stream using Apache Airflow.
• Load: Maintained a ML/AI data warehouse composed of various Google Cloud technology: Cloud Storage buckets for raw data storage, Big Query data warehouse for storing data transformation from Apache Airflow, and persistent cloud disk drives for high throughput access to terabytes of training data.

• Designed and administered astronomy research MySQL database storing over 4 million calibrated supernovae images and metadata to replace the old method of searching for supernovae images using a file server.
• Designed and developed API web service (Python Flask framework) to interface the mentioned database. This will be used in the frontend to access the database.
• Developed an XML parser for astronomy's VOEvent convention. This was used in multimessenger astronomy to parse NASA's Gamma-ray Coordinates Network alerts (e.g., gravitational wave alerts). This was the first step in the process that enrolled our team into an exciting field of multimessenger astronomy.
• Deployed, automated, and administrated research pipelines and softwares in Linux production server using Git for version control, Make + Gunicorn + Apache + systemd for deploying, and automating and monitoring web services.

• Developed regression models relating CCD measurements to solar irradiance. This relation is then used in an ARIMA model to forecast solar irradiance which is important for PV grid management. The models provided a minimal working product to further assess the potential of all-sky cameras as a useful imaging tool for solar irradiance forecasting.