// PhD Researcher · Software Engineer
Hamza
Attarwala
Building intelligent systems at the intersection of software engineering and machine learning. PhD student at Polytechnique Montréal, working on scalable AI applications.
// PhD Researcher · Software Engineer
Building intelligent systems at the intersection of software engineering and machine learning. PhD student at Polytechnique Montréal, working on scalable AI applications.
About
I'm a PhD student in Computer Engineering at Polytechnique Montréal, at the Software Emerging Technologies (SæT) Lab under the supervison of Dr. Mohammed Hamdaqa. I completed my MSc. in Applied Modeling & Quantitative Methods at Trent University, under the supervision of Dr. Quazi Rahman.
My research sits at the intersection of Artificial Intelligence and Software Engineering (AI4SE), with a focus on harnessing Large Language Models (LLMs) to augment and automate core SE tasks. Currently, I'm investigating how LLMs can improve areas such as requirements engineering, code generation, and formal specification — with the broader goal of making software development more reliable, efficient, and accessible.
My work bridges rigorous research and practical engineering — from building EV charging platforms to designing AI-powered course recommenders!
Previously at EasyFits and EVDrop, I led backend development, optimized APIs, and deployed scalable cloud services. I also spent four years as a Teaching Assistant at Trent University, helping students in learning Python, Java, and Linux workflows.
Education
Research
Large Language Models (LLMs) are increasingly used to generate Object Constraint Language (OCL) constraints from natural language specifications and UML class diagrams. However, existing work mainly focuses on improving accuracy, with limited understanding of why these models fail. This study investigates the underlying causes of LLM failures in OCL generation, framing the task as a graph reasoning problem over UML class diagrams. We conduct an empirical evaluation using the PathOCL dataset across six state-of-the-art LLMs. We analyze the impact of UML structural properties (e.g., navigation depth and model complexity), lexical similarity, prompt ordering strategies, and graph-aware prompting on OCL correctness. We find that OCL generation performance significantly degrades with increasing navigation depth and structural complexity. Lexical similarity has limited influence, while textual ordering of UML elements affects performance. Graph-based prompting yields partial improvements but does not eliminate structural reasoning errors. OCL generation is primarily constrained by graph reasoning limitations rather than purely linguistic factors. These results highlight structural reasoning as a key bottleneck for current LLMs in model-driven engineering tasks.
Large Language Models (LLMs) have shown promising performance in generating Object Constraint Language (OCL) specifications from natural language requirements. However, existing evaluations often rely on publicly available UML models, which may overstate generalization beyond familiar lexical and structural patterns. This paper introduces a transformation-driven benchmarking approach for OCL generation, based on deterministic, semantics-preserving UML model transformations—such as identifier renaming, and attribute and association reification—that preserve specification intent while invalidating syntactic cues. We evaluate a diverse set of closed-source and open-source LLMs of varying scales under zero-shot, few-shot, and chain-of-thought prompting, measuring syntactic accuracy (well-formed and type-correct OCL) and semantic accuracy (correct interpretation with respect to the UML model). While leading models achieve high performance on non-transformed models—up to 92.10% syntactic accuracy and 74.56% semantic accuracy—performance degrades substantially under transformation, with best semantic accuracy dropping to 55.96% for closed-source models and 42.20% for open-source models. These results demonstrate that current LLMs remain brittle under semantically equivalent but structurally altered models, revealing a reliance on syntactic patterns rather than robust model-aware reasoning, and underscore the need for transformation-based benchmarks in evaluating formal specification generation.
Predicting vehicle speed at critical road segments, such as pedestrian crossings during left-turn maneuvers at signalized intersections, is essential for improving traffic safety and supporting autonomous driving systems. This thesis presents a novel two-stage hybrid deep learning framework enhanced with reinforcement learning to forecast vehicle left-turn speed at pedestrian crossings.
Experience
EVDrop ↗ · Remote
EasyFits ↗ · Remote
Trent University · Peterborough, ON
Contact
Whether it's research collaboration, a project idea, or just a conversation — my inbox is open.