Data Intensive Software Analytics (DISA) Lab

Director and Founder: Dr. Gias Uddin

The DISA Lab conducts research on the trustworthiness assessment and improvement of AI-powered systems, with a focus on robustness, safety, and security. We employ statistical methods, data science, and software engineering principles to develop techniques that make AI assistants more reliable and secure. Our mission is to enhance the productivity and effectiveness of knowledge professionals—including data scientists and software engineers—through robust and trustworthy AI tools.

Our research spans four interconnected areas:

Engineering Trustworthy AI Assistants

We develop systematic testing and verification techniques to assess and improve the reliability of large language models and AI assistants. Our work includes detecting and mitigating hallucinations, incorrectness, and bias through metamorphic relations and other rigorous methodologies. We create frameworks that evaluate the fairness and safety of AI systems, ensuring they perform dependably and transparently in real-world applications.

AI for Software Issue Management

We leverage machine learning and deep learning to automate critical software engineering tasks, including bug triage, crash deduplication, and technical documentation analysis. Our research employs transformer-based models to accelerate software maintenance workflows and enhance the accuracy of issue classification and resolution.

AI for Coding and Security

We investigate how large language models can assist in code generation, reasoning, and validation. Our work includes developing techniques for loop invariant generation, improving code quality through metamorphic testing strategies, and creating intelligent agents that autonomously identify and resolve software engineering defects. We also study the security implications of AI-generated code.

Quality and Security of Crowd Technical Knowledge

We analyze the quality, security, and trustworthiness of technical knowledge shared in community-driven platforms like Stack Overflow. Our research encompasses API usage pattern extraction, mining developer sentiment and best practices, identifying security vulnerabilities in crowd-sourced code examples, and understanding reputation dynamics that shape technical communities.

Current Openings at the DISA Lab

We are seeking passionate and self-motivated researchers to join our team. Ideal candidates demonstrate strong technical foundations in software engineering and deep learning, combined with a commitment to rigorous research.

PhD Position

We welcome applications from prospective PhD students with solid backgrounds in software engineering, machine learning, and/or human-computer interaction. You will work on cutting-edge problems spanning our research areas and develop expertise in rigorous software engineering and deep learning techniques. A background in machine learning fundamentals is highly valued. If interested, please send your CV, academic transcripts, and a list of relevant publications or projects.