Farzana Yasmin Ahmad

PhD Candidate in Computer Science

University of Virginia - Biocomplexity Institute

About Me

I am a PhD candidate in Computer Science at the Biocomplexity Institute, University of Virginia, advised by Professor Geoffrey C. Fox. My research focuses on developing and evaluating machine learning methods for complex scientific problems, with particular emphasis on generative AI for physics simulations.

Expected graduation: Spring 2027

Education

Ph.D. in Computer Science

University of Virginia — Biocomplexity Institute

Advisor: Professor Geoffrey C. Fox

2021 – Present (Expected: Spring 2027)  |  CGPA: 3.90 / 4.00

Master of Computer Science (M.C.S.)

University of Virginia — Department of Computer Science

Spring 2026  |  CGPA: 3.90 / 4.00

B.Sc. in Computer Science and Engineering

Bangladesh University of Engineering and Technology (BUET)

2015 – 2019  |  CGPA: 3.69 / 4.00

Research Interests

Scientific ML for Complex Systems

Applying machine learning to scientific computing challenges

Generative AI

Diffusion models, flow matching, VAEs, and fine-tuning techniques

Model Evaluation

Rigorous evaluation of generative models for fidelity and reliability

Technical Skills

Python PyTorch TensorFlow Hugging Face Diffusion Models Flow Matching Multi-GPU Training HPC Git/GitHub LLM Systems RAG Pipelines

Selected Publications

CaloBench: A Benchmark Study of Generative Models for Calorimeter Showers
Ahmad, F.Y., Venkataswamy, V., and Fox, G. - International Symposium on Benchmarking, Measuring and Optimization (Accepted - to appear), Springer, 2024.
Distributed Principal Component Analysis for Real-time Big Data Processing
Meem, J.A., Ahmad, F.Y., and Adnan, M.A. - Proceedings of the 7th International Conference on Networking, Systems and Security, 2020, pp. 89–99.

Featured Projects

LANTERN: Physics-Guided Diffusion for Calorimeter Showers

Developed a novel physics-guided diffusion model augmented with auxiliary training objectives including correlation-aware CFD, MVN likelihood, and energy consistency constraints. Implemented timestep weighting and warm-up strategies for improved training.

View on GitHub →

CaloBench: Systematic Benchmark of Generative Models

Conducted a comprehensive multi-detector evaluation of generative models for calorimeter data. Introduced the Correlation Frobenius Distance (CFD) metric to quantify preservation of second-order structure and analyzed energy-stratified behavior.

View on GitHub →

LLM-Based Forecasting of Crisis-Induced Migration

Systematically evaluated GPT-5, Gemini 2.5 Flash, and Claude Sonnet 4.5 for forecasting temporal return migration patterns following the 2022 Russian invasion of Ukraine. Assessed forecasting accuracy, robustness, and reasoning processes using real-world migration, conflict, and historical data. Findings indicate LLMs require human oversight for reliable crisis forecasting.

View on GitHub →

PhinD-me-if-You-Can

Built an AI-powered tool during the Claude for Good 2025 hackathon (7.5 hours) that matches students with potential PhD advisors using LLM APIs. Implemented quantitative match scoring with qualitative explanations using Streamlit.

View on GitHub →

Calo-RAG: Literature Retrieval for Calorimeter Modeling

Built a RAG pipeline for practicing LLM+IR techniques with prompt templates, citation-aware retrieval, and basic reranking. Evaluated coverage using precision/recall metrics on curated queries.

View on GitHub →

Awards & Recognition

2026 - Best Flash Talk Presentation (Graduate Category), ACM Capital Region Celebration of Women in Computing (CAPWIC)
2025 - Lightning Talk, UVA CS Research Symposium
2025 - Poster Presentation Finalist, UVA Engineering Research Symposium (UVERS)
2023 - Invited to CRA-WP Graduate Cohort Workshop for Women
2019 - Dean's List Scholarship, Bangladesh University of Engineering and Technology
2017, 2018 - University Merit List Scholarship, BUET

Teaching Experience

Graduate Teaching Assistant - University of Virginia

Lecturer (2019-2020)

Department of Computer Science, United International University

Service & Leadership