02 / Research topics
Potential PhD Research Topics in Software Engineering
Explore suggested topics and the methods, tools and expertise that could support each investigation.
1
Trustworthy Generative Artificial Intelligence for Requirements Engineering and Software Maintenance
Rapid adoption of large language models is transforming requirements and maintenance, yet hallucinations, hidden bias, weak traceability, and insecure generated changes limit dependable use. Existing benchmarks rarely measure contextual correctness and long-term maintenance effects together. This research could develop human-guided generation, evidence-linked outputs, risk-aware validation, and repository-grounded agents. Objectives include defining trustworthiness metrics, creating benchmark datasets, comparing collaboration strategies, and evaluating productivity, defect rates, explainability, and developer confidence through controlled and industrial studies.
Applications and use cases
2
Autonomous Self-Healing Software Systems for Resilient Cloud-Native Applications
Despite advanced observability, cloud systems still experience cascading failures that demand costly human diagnosis and intervention. Current self-healing approaches are often rule-based, reactive, and difficult to validate under novel failures. This proposal could combine causal diagnosis, safe reinforcement learning, and automated remediation with explicit recovery guarantees. Objectives are to model failure propagation, generate explainable recovery plans, constrain unsafe actions, and assess resilience improvements using realistic fault injection, production traces, and human-operated baselines.
Applications and use cases
3
Carbon-Aware Software Engineering for Sustainable Cloud and Edge Computing
Software decisions increasingly influence energy consumption and carbon emissions, but developers lack actionable, accurate guidance across architectures and deployment regions. Existing green practices frequently optimize performance or cost without considering workload variability, embodied emissions, and user experience. This research could create carbon-aware architectural patterns, measurement models, and automated optimization tools. Objectives include quantifying trade-offs, integrating emissions into engineering workflows, validating adaptive scheduling strategies, and producing practical guidelines through experiments on cloud and edge workloads.
Applications and use cases
4
Engineering Reliable and Verifiable Software for Quantum Classical Hybrid Applications
Quantum software development remains immature, with unstable hardware, limited testing techniques, probabilistic behavior, and difficult debugging creating major reliability gaps. Existing methods often adapt classical practices without addressing quantum-specific semantics and hardware noise. This project could establish quality models, metamorphic testing strategies, resource-aware compilers, and hybrid verification workflows. Objectives are to characterize failure modes, develop automated test generation, assess portability across quantum backends, and validate the approach using representative optimization, simulation, and machine-learning applications.
Applications and use cases
5
Privacy-Preserving Software Analytics Across Distributed Development Organizations
Organizations possess valuable engineering data but cannot freely share repositories, developer activity, or defect information because of privacy, intellectual property, and employment concerns. Consequently, software analytics models often suffer from limited generalizability and reproducibility. This research could combine federated learning, differential privacy, and secure aggregation for cross-organization prediction without exposing sensitive data. Objectives include measuring privacy-utility trade-offs, detecting dataset bias, improving transferability, and evaluating predictions for defects, delivery risks, and developer well-being.
Applications and use cases
6
Human-Centered Developer Experience Engineering for AI-Augmented Software Teams
AI coding assistants can increase speed while also introducing cognitive overload, skill atrophy, accountability ambiguity, and coordination problems. Current productivity studies emphasize output volume rather than comprehension, inclusion, satisfaction, and long-term quality. This proposal could develop a multidimensional developer experience framework and adaptive assistant policies based on task context and individual needs. Objectives include identifying harmful interaction patterns, designing transparent interventions, measuring team-level outcomes, and validating them through field deployments and longitudinal mixed-method studies.
Applications and use cases
7
Secure-by-Design Software Supply Chains Through Continuous Provenance and Risk Analysis
Modern software depends on opaque third-party packages, automated pipelines, and shared artifacts, enabling dependency confusion, malicious updates, and compromised build systems. Existing security tools generate fragmented alerts and rarely connect provenance with architectural and business risk. This research could create an evidence-based supply-chain assurance model integrating provenance graphs, behavioral analysis, policy enforcement, and developer feedback. Objectives are to detect novel attacks, prioritize actionable risks, evaluate false-positive reduction, and test adoption across realistic continuous integration environments.
Applications and use cases
8
Digital Twin Driven Verification and Evolution of Cyber-Physical Software Systems
Cyber-physical systems are difficult to test because physical environments are expensive, hazardous, variable, and only partially observable. Digital twins offer scalable experimentation, but fidelity gaps and synchronization errors can invalidate engineering conclusions. This research could develop uncertainty-aware twins that combine runtime data, simulation, and formal constraints for continuous verification. Objectives include defining fidelity measures, detecting divergence, generating realistic scenarios, and evaluating safety, maintainability, and deployment decisions in industrial and autonomous systems.
Applications and use cases
9
Inclusive Software Engineering Methods for Accessible and Culturally Adaptive Digital Products
Many software products technically meet minimum accessibility standards while still excluding users through poor interaction, language assumptions, cultural bias, or inaccessible development processes. Existing engineering methods rarely integrate lived experience throughout requirements, design, testing, and maintenance. This research could establish participatory, AI-supported accessibility practices with measurable inclusion outcomes. Objectives include identifying overlooked barriers, developing inclusive requirements and test techniques, evaluating automated assistance for bias and accessibility, and demonstrating impact in public, educational, and healthcare applications.
Applications and use cases
10
Evidence-Based Continuous Software Engineering Education Using Intelligent Learning Environments
Software engineering education struggles to keep pace with AI tools, secure development, distributed teamwork, and rapidly changing technologies. Traditional assessments often reward code completion without measuring reasoning, collaboration, maintainability, or ethical judgment. This research could create intelligent learning environments that provide evidence-based feedback while preserving academic integrity and learner agency. Objectives include modeling professional competencies, generating authentic project assessments, evaluating personalized interventions, and validating learning transfer through classroom studies and graduate workplace follow-up.
Applications and use cases