Saima Afrin
PhD Candidate
Department of Computer Science
William & Mary
251 Jamestown Rd., Williamsburg
Virginia, USA
I am currently working under the supervision of Dr. Antonio Mastropaolo as a member of the AURA Lab, where I contribute to research on AI4SE with a focus on optimizing the performance and adaptability of code-focused AI models.
Before starting my Ph.D., I earned a Bachelor of Science in Computer Science and Engineering from Daffodil International University, graduating in the top tier of my class. My early research explored supervised and deep learning techniques across a variety of real-world problems, contributing to several peer-reviewed publications.
Alongside my academic journey, I also served as a full-time Lecturer at Daffodil International University, where I mentored undergraduate students and contributed to a collaborative teaching and research environment. I am passionate about impactful, collaborative research and regularly engage in conferences, seminars, and innovation events to exchange ideas with the broader scientific community.
news
Our paper "Quantize with Confidence? An Empirical Study of Quantization for Code Generation" has been accepted at ICSME 2026, the 42nd IEEE International Conference on Software Maintenance and Evolution!
Our paper "Large Language Models for Code Generation from Multilingual Prompts: A Curated Benchmark and a Study on Code Quality" is under review at ACM TOSEM.
Started my summer research internship at the University of Sannio, Italy, under the supervision of Prof. Massimiliano Di Penta (June–September 2026).
Serving as a Program Committee member for the Replication and Negative Results (RENE) Track of ICSME 2026.
Served as a Program Committee member for the CAIS 2026 Workshop on Agentic Software Engineering (AgenticSE).
Serving as a reviewer for leading software engineering journals — Empirical Software Engineering (EMSE), ACM TOSEM, and IEEE TSE.
Pleased to serve as a PC member for the Mining Challenge Papers track at MSR 2026.
Our paper "Parameter-Efficient Multi-Task Fine-Tuning in Code-Related Tasks" is under review at ACM TOSEM. Preprint: arXiv:2601.15094.
Our paper "Evaluating the Impact of Post-Training Quantization on Large Language Models for Code Generation" has been accepted in the Research track of ICPC 2026, the 34th IEEE/ACM International Conference on Program Comprehension!
Honored to serve as a Junior PC member at MSR 2026 (23rd International Conference on Mining Software Repositories).
Our paper "A Systematic Literature Review of Parameter-Efficient Fine-Tuning for Large Code Models" has been accepted for publication in ACM Transactions on Software Engineering and Methodology (TOSEM)!
Served as a reviewer at "Journal of Systems and Software" (JSS) .
Presented our paper "Is Quantization a Deal‑breaker: Empirical Insights from Large Code Models" at ICSME 2025 Conference .
I got the opportunity to serve as a student volunteer at the ICSME 2025 Conference, took place in Auckland, NewZealand .
Pleased to share that I served as a PC member at the "International Workshop on Analytics for Software Product and Process Improvement 2025" (A-SPPI 2025), co-located with PROFES 2025 .
I was awarded the ICSME 2025 NSF Student Travel Grant.
Our paper "Is Quantization a Deal‑breaker: Empirical Insights from Large Code Models" was accepted as a Full Paper at ICSME 2025 .
Our study on Code LLMs and Quantization "Quantizing large language models for code generation: A differentiated replication" is available on arXiv:2503.07103.
Presented our paper "Resource‑efficient & effective code summarization" at at Forge (AI Foundation Models & Software Engineering) 2025 conference, co-located with ICSE 2025 (Ottawa, Canada).
Our paper "Single-GPU GNN systems: Traps and Pitfalls" has been accepted at USENIX 2024.
Our paper "Resource‑efficient & effective code summarization" was accepted at Forge (AI Foundation Models & Software Engineering) 2025.