PhD Research

Research Projects

My dissertation and applied research spans probabilistic machine learning, astronomical image analysis, and high-performance CT reconstruction, united by a focus on reliability, uncertainty, and scalability in high-dimensional data.

01
Bayesian ML · Thesis

Trustworthy Variational Bayesian Inference


Variational Inference Logistic Regression Calibration HPC Parallel Uncertainty

Trustworthy Variational Bayesian Inference for Reliable Machine Learning

Developed a scalable Bayesian machine learning framework that improves the reliability of predictive models by addressing a key limitation of standard variational inference: the systematic underestimation of uncertainty.

The Trustworthy Variational Bayesian (TVB) approach for logistic regression remains robust under model misspecification while maintaining computational efficiency. By combining principled statistical modeling with high-performance optimization, the method produces better-calibrated uncertainty estimates and more stable predictions in high-dimensional settings.

This research directly supports the development of trustworthy AI systems, where understanding confidence levels and failure modes is critical for real-world decision-making in national security and applied ML contexts.
02
Astrostatistics · Thesis

Multi-Band Astronomical Image Reconstruction


Nonparametric Inference Signal-to-Noise Multi-sensor Fusion Covariance Estimation

Multi-Band Statistical Reconstruction for Low-Signal Astronomical Imaging

Designed a nonparametric statistical framework for reconstructing faint astronomical signals by jointly analyzing multi-wavelength imaging data. The approach leverages cross-band correlations between continuum and emission-line observations to improve background estimation and enhance detection in low signal-to-noise conditions.

By integrating spatial modeling, local covariance estimation, and controlled signal-to-noise experiments, this work demonstrates significant improvements over traditional single-band methods.

The methodology generalizes to multi-sensor data fusion problems, offering broader relevance to any application that requires extracting weak signals from noisy, high-dimensional data: surveillance, geospatial imaging, biomedical sensing.
03
Applied Research · Sandia

CT Reconstruction & Imaging Pipelines


GPU Acceleration Anomaly Detection End-to-End Pipelines Benchmarking
Sandia National Laboratories

Scalable CT Reconstruction and Analytics Pipelines for High-Dimensional Imaging

Developed high-performance data pipelines and GPU-accelerated algorithms for large-scale computed tomography reconstruction at Sandia National Laboratories. This work focused on building end-to-end systems for ingesting, processing, and analyzing high-dimensional imaging data with an emphasis on performance, reliability, and reproducibility.

Through systematic validation, benchmarking, and anomaly detection, the pipeline enables efficient processing of complex datasets while identifying failure modes in real-world inspection systems.

These contributions support mission-critical nondestructive inspection applications where robust, scalable imaging analytics are essential, directly aligned with national security R&D at Sandia Labs where I have served as a year-round intern since June 2021.