THREADS Lab is the research group of Naser Ezzati-Jivan, Associate Professor in the Department of Computer Science at Brock University. Our research focuses on software observability, tracing, performance engineering, trustworthy and agentic AI systems, and dependable heterogeneous systems.

We welcome inquiries from master’s and PhD students, visiting researchers, postdoctoral fellows, and collaborators whose interests align with these areas. Please use the Contact page to get in touch.

Research directions

A system can only be understood, trusted, or improved to the extent that its runtime behaviour can be observed as evidence — so we build the tracing and analysis methods that turn what a system actually does at runtime into evidence you can reason about, whether that system is a distributed service, an AI agent, or a robot.

Our work sits in three connected pillars. Current projects are listed under each; some pillars also include broader standing directions that are not tied to a single named project.

1. Software observability, tracing & performance engineering

Adaptive, multilevel tracing and root-cause methods spanning application code down to kernel and hardware evidence, plus distributed, multicore, cloud, and virtualized systems more broadly.

  • Multi-Level Adaptive Tracing for Enhancing Software Reliability and Performance — Mitacs Business Strategy Internship (BSI), with Ciena Corporation. Naser Ezzati-Jivan, academic supervisor (PI).
  • Observability Compensation Paradigm: Leveraging Adaptive Execution Tracing and Analysis — NSERC Discovery Grant. Naser Ezzati-Jivan, PI.

2. Trustworthy & agentic AI systems

Making AI-driven and agentic software systems observable, auditable, and fair, including EDI-by-design approaches to inclusive AI research practice.

  • EDI-by-Design in the Age of AI: A Co-Created Framework and Decision Support Tool for Inclusive Research Planning — NSERC Dimensions Canada, SSHRC Insight Development, and Match of Minds. Naser Ezzati-Jivan, PI, in partnership with the University of Ottawa.
  • RapidBOM: Multimodal Document AI and Privacy-Preserving Knowledge Graphs for Engineering Sourcing — Mitacs Accelerate, with Interconnect Dynamics. Naser Ezzati-Jivan, academic supervisor (PI).
  • Intelligent Software Agents for Autonomous Monitoring and Repair — Mitacs Globalink Research Internship.
  • Harnessing Large Language Models for Automated Software Log Understanding — Mitacs Globalink Research Internship.

3. Heterogeneous execution & embodied/physical systems

Extending observability beyond conventional software boundaries into heterogeneous hardware and the physical world; this pillar is the newest and most exploratory of the three.

  • Observability and benchmarking for AI inference on heterogeneous hardware — ongoing project.
  • Observability and energy-usage analysis for physical AI platforms, including robots, drones, and other embodied systems.

Completed project:

How we work

We are an empirical, systems-first group: claims are backed by measurements on real workloads, and methods are evaluated against production-scale traces wherever possible. Much of our work ships as open-source artifacts — several papers have accepted artifacts with persistent identifiers — and our publication catalog records the evidence level behind every summary. We collaborate closely with industry (Ciena, Ericsson, and others) so that research problems come from, and return to, systems that people actually run.

Team

THREADS Lab currently includes PhD and MSc students, undergraduate researchers, research assistants, and Mitacs Globalink interns. See the People page for the full roster of current members and alumni.

Explore our research publications and contributions to the field. The catalog provides technical summaries, identifiers, and citation formats for every paper.

View all publications