About Me

I am a PhD candidate at Nanda Lab, University of Ottawa under the supervision of Prof. Lionel Briand. My main research interests include all stages of system engineering for complex AI-enabled systems including robots, autonomous vehicles, and cyberphysical systems. This encompasses system design and modeling, requirements engineering, quality assurance, testing, validation, and runtime monitoring. During my PhD, I have developed novel methods for automated simulation-driven testing and monitoring. These methods leverage a variety of metaheuristics and machine learning techniques, and have been successfully applied to real-world systems in autonomous driving and aviation.

Education

PhD in Digital Transformation and Innovation (AI System Safety)

2020 - present
University of Ottawa

MSc in System Science

2018 - 2020
University of Ottawa

BSc in Aerospace Engineering

2012 - 2017
Sharif University of Technology

Experiences

AI Engineer Intern

Aug 2023 - Mar 2024
Quantum Technologies Inc., Waterloo ON

PhD Research Intern

May 2021 - Feb 2024
Auxon Technologies, Ottawa ON

System Safety Research Intern

Jan 2020 - May 2020
Brane Inc, Ottawa ON

Research

My research focuses on making robots and cyberphysical systems safer. Specifically, I am interested in Machine Learning (ML) components and how they impact the safety of the system. To that end, I have developed an effective and efficient tool to identify the hazard boundary of ML components via scalable parallel simulations. Furthermore, I have developed methods and tools to monitor the ML component during its operation at runtime. My research is supported by NSERC Canada Research Chairs program, Auxon Corporation, Mitacs and Ontario Graduate Scholarship.

Runtime Safety Monitoring of Learned Components - In this project, I developed safety monitors for learned components, using state-of-the-art DL-based probabilistic forecasting models. I have also performed extensive empirical evalaution in terms of prediction accuracy and runtime performance.
Hazard Boundary Identification of ML-enabled Autonomous Systems - In this project we proposed MLCSHE, a cooperative coevolution algorithm with a probabilistic fitness function to identify the hazard boundary of a machine learning component which is embedded in an atuonomous system. MLCSHE uses the actual system and scalable parallel simulations to identify the hazard boundary.

Publications

  • System Safety Monitoring of Learned Components Using Temporal Metric Forecasting
  • Sepehr Sharifi, Andrea Stocco, Lionel C. Briand
    ArXiv Preprints
  • Identifying the Hazard Boundary of ML-enabled Autonomous Systems Using Cooperative Co-Evolutionary Search
  • Sepehr Sharifi, Donghwan Shin, Lionel C. Briand, Nathan Aschbacher
    Transactions on Software Engineering, vol. 49, no. 12, pp. 5120-5138, 2023
  • Specification and analysis of legal contracts with Symboleo
  • Alireza Parvizimosaed, Sepehr Sharifi, Daniel Amyot, Luigi Logrippo, Marco Roveri, Aidin Rasti, Ali Roudak, John Mylopoulos
    Journal of Software and System Modeling, 21, 2395–2427, 2022
  • Symboleo: Towards a specification language for legal contracts
  • Sepehr Sharifi, Alireza Parvizimosaed, Daniel Amyot, Luigi Logrippo, John Mylopoulos
    28th IEEE International Requirements Engineering Conference, 2020
  • Subcontracting, assignment, and substitution for legal contracts in Symboleo
  • Alireza Parvizimosaed, Sepehr Sharifi, Daniel Amyot, Luigi Logrippo, John Mylopoulos
    39th International Conference on Conceptual Modeling, 2020
  • Social Dependence Relationships in Requirements Engineering
  • John Mylopoulos, Daniel Amyot, Luigi Logrippo, Alireza Parvizimosaed, Sepehr Sharifi
    13th International i* Workshop, 2020
  • Goal Modeling for FinTech Certification
  • Sepehr Sharifi, Patrick McLaughlin, Daniel Amyot, John Mylopoulos
    13th International i* Workshop, 2020