CV

Profile

Postdoctoral Fellow at Toronto Metropolitan University and Vector Faculty Affiliate Researcher, working on trustworthy machine learning, certified robustness, and sequential change detection. I develop mathematical methods and reproducible Python/PyTorch experiments, with industry experience in product experimentation, causal analysis, and AI-agent evaluation.

Download my CV (PDF). This two-page version summarizes my research and applied AI experience.

Education

• Ph.D., Computer Systems Engineering, Toronto Metropolitan University, September 2021–May 2026. Research area: neural-network robustness verification. Thesis: Scalable Robust Verification of Neural Networks. Supervisor: Prof. Reza Samavi. GPA: A+.
• M.Sc., Electrical Engineering (Integrated Circuit Engineering), University of Tehran, 2017–2020. Thesis: Phase Array Inductive Wireless Power Transfer for Moving Load. Supervisor: Prof. Shahin Jafarabadi Ashtiani. GPA: 17.29/20 (3.7/4.0); thesis evaluation: Excellent.
• B.Sc., Electrical Engineering (Communication), Isfahan University of Technology, 2013–2017. Capstone: Basics and Applications of Digital Controlled Oscillators (supervisor: Prof. Masood Omoomi). Elective project: Design and Simulation of an X-Band Microwave Amplifier (supervisor: Prof. Abolghasem Zeidaabadi Nezhad). GPA: 16.66/20 (3.67/4.0).

Employment history

• Postdoctoral Fellow, Trustworthy AI Research Lab, Toronto Metropolitan University, June 2026–present. Develop scalable certified-training methods and analyze LP, SOCP, and SDP relaxations for neural-network robustness.
• Industry Research Collaborator, jamais51 Technologies Inc. and Toronto Metropolitan University, June 2026–present. Formalize threat models and integrity conditions and test an independent settlement verifier.
• Head of Product & Data, ParagraphAI, June–September 2026. Designed controlled product experiments, analyzed user and growth data, and configured AI-agent workflows. SpeechLP was a project within this role; collaborated with speech-language pathologists and developers on product planning, content verification, and speech-therapy games.
• Doctoral Researcher / Research Assistant, Trustworthy AI Research Lab, Toronto Metropolitan University, September 2021–May 2026. Developed robustness certificates, cascading verification, and research software.
• Master’s Researcher / Research Assistant, Analog Integrated Circuit Design Laboratory, University of Tehran, 2017–2020. Modeled inductive wireless power systems and contributed to a deep-brain-stimulator project.

Research affiliation: Vector Faculty Affiliate Researcher.

See Experience for responsibilities and dates.

Selected projects

• Scalable neural-network robustness verification: Developed sparse SOCP relaxations and the CORE class-wise verification cascade. On a fully connected MNIST benchmark, CORE matched SDP-certified robust accuracy while reducing average verification time by up to 86.30%.
• CoRe-CROWN: Combined a Dual-CROWN training objective with sparse SOCP-derived regularization, influence-based block selection, and local lifted constraints; evaluated complete certified accuracy.
• DI-SCUSUM: Co-developed training-free streaming change detection using closed-form Gaussian-smoothed scores. In a calibrated anisotropic Gaussian experiment, achieved approximately 91% lower detection delay than score-based CUSUM at matched false-alarm levels.
• Secure Machine Learning / UN PET Lab Hackathon: Applied feature screening and SmartNoise queries to estimate hidden labels under a privacy budget. Logistic regression achieved a 0.6934 competition score at total epsilon expenditure of 1.85. Project repository.

See Research and Selected Projects for methods, benchmarks, and earlier engineering projects.

Publications and manuscripts

Manuscripts under review

• M. Maleki, D. Sadig, R. Sidibomma, M. Fazlyab, A. Adibi, and R. Samavi. Certified Robustness via Sparse Second-Order Cone Relaxations. Under review, AISTATS 2027.
• A. Adibi*, M. Maleki*, S. Kulkarni, and H. V. Poor. Quickest Change Detection with Diffusion-Integrated Scores. Under review, AISTATS 2027. *Equal contribution.
• M. Maleki, D. Sadig, A. Adibi, and R. Samavi. CoRe-CROWN: Conic-Regularized Certified Robust Training. Under review, AAAI 2027.

Peer-reviewed publications

See the publications page for the SaTML, Canadian AI, and journal papers, summaries, and download links.

Teaching and mentorship

Course instructor: ELE 202 — Electric Circuit Analysis, Toronto Metropolitan University (Winter 2025).

Course graduate assistant, Toronto Metropolitan University: MTH 410 — Statistics (Winter 2026); MTH 380 — Probability and Statistics I (Fall/Winter 2023); MTH 312 — Differential Equations and Vector Calculus (Fall 2024); MTH 110 — Discrete Mathematics I (Fall 2024); ELE 504 — Electronic Circuits II (Fall 2024); ELE 404 — Electronic Circuits I (2022–2026); ELE 302 — Electric Network (2021–2025); ELE 202 — Electric Circuit Analysis (2022–2026).

Course graduate assistant, University of Tehran: Electronics III (Fall 2018); Electronics II (2018–2019); Data Converters (Spring 2019).

Research mentorship: Assisted with the research supervision of Rushendra Sidibomma (Mitacs Globalink, 2023), Daniel Sadig (M.A.Sc., graduated 2026), and Hao Luo (M.Eng., graduated 2024). See Teaching and Mentorship for their projects and collaborations.

Academic and professional service

• Invited Reviewer, Machine Learning (Springer Nature), 2026.
• Invited Reviewer, 18th Asian Conference on Machine Learning (ACML), 2026.

Honors and awards

• TMU Graduate Fellowship (2023–2025), Graduate Development Award (2021–2024), FEAS Graduate Funding (2021–2023), and International Student Scholarship (2022–2023).
• Top 0.5% of 40,000 candidates, Nationwide Graduate Qualifying Examination in Electrical Engineering, 2017.
• Top 1% of 251,956 candidates, Nationwide University Entrance Qualification Test, 2013.

See Service and Awards for the full list.

Technical skills

• Programming and ML: Python, PyTorch, TensorFlow, NumPy, pandas, SciPy, scikit-learn; SQL, MATLAB, and C.
• Research methods: Convex optimization, LP, SDP, SOCP, bound propagation, robust training, statistical inference, controlled experiments, causal inference, change detection, and model evaluation.
• Tools: Git, GitHub, Linux, Jupyter, and LaTeX.