Biography

I am a 3rd Year PhD student and a mom of two. Love research, reading books and hanging out with my family.

Academic Status

PhD Student - 3rd

Research Area/Department

Computer Science; Data Science; Machine Learning/AI

Major/Specialty

Computer Science

Degrees Earned or in Progress

PhD 3rd Year

Academic Preparation

Advanced Machine Learning and Pattern Recognition, Human-Centric Deep Learning, Scalable Systems for Supercomputing, High Performance Computing, Seminar on Quantitative Analysis

Research/Publications

My research so far has been published in ICASSP 25, IEEE Compsac 24 and I have an accepted paper as a second author at NeurIPS.

Research/Academic Interests

My research interests lie at the intersection of causal modeling and generative modeling, with a particular focus on applications in High-Performance Computing (HPC). My first paper (published in IEEE Compsac 24) aimed at creating realistic synthetic traces that preserve causal dependencies and performance dynamics as many HPC workloads are under-represented or noisy, making it difficult to train robust predictive or causal models. I did a comparison between the quality of data generated by a GAN based model and an LLM-based model along with introducing a metric to compare the quality of synthetic data. My other paper published in ICASSP 25 was based on improving the results of the synthetic data via providing additional semantic information to the LLM during the training process. Currently, I am working on a decision-support framework that combines causal inference and generative modeling to provide actionable insights for HPC system management.

Computational and Data Science Areas

Artificial Intelligence and Intelligent Systems; Computer Science

Motivation

High-performance computing (HPC) systems generate vast amounts of trace and performance data that capture the intricate interactions between hardware, software, and workloads. Yet, despite their richness, these datasets remain understudied in terms of structure. These datasets represent complex, interdependent systems where standard statistical or predictive methods often mask the true drivers of performance. At the same time, causal modeling remains underutilized in HPC, even though it offers the tools to disentangle dependencies, identify key bottlenecks, and ask meaningful “what-if” questions. This gap between the complexity of HPC data and the limited application of causal approaches strongly motivates my research. By integrating causal modeling with generative methods, I aim to better understand these systems, uncover actionable insights, and ultimately improve performance prediction, scheduling, and energy efficiency in large-scale computing environments.

Lightning Talk Title

Causality-Aware AI Makes Complex Decisions Explainable and Trustworthy

Keywords (Maximum 20 words)

Causal Analysis; causal inference; mediation analysis; scheduling; representation learning; augmentation; transfer learning;