Physical AI &
connected systems
Bringing sensing, connectivity, learning, and control together for intelligent systems operating under uncertainty.
AI researcher · Montreal, Canada
Mehran Attar.
Intelligence for
real-world systems.
I work at the intersection of artificial intelligence, optimization, and control—developing methods that make complex systems more capable, efficient, and reliable.
Currently
AI Researcher/Ericsson Canada
Across research
and industry
01 / Research directions
From learning
to reliable action.
My research connects data-driven methods with the constraints of physical and connected systems: uncertainty, safety, scale, and efficiency.
Physical AI &
Foundation models
for telecom
Adapting large-scale models to network and sensing data through representation learning and multimodal reasoning.
Quantum-inspired
optimization
Designing efficient methods for training, model selection, and decision-making in high-dimensional AI and telecom problems.
Secure, data-driven
control
Developing control architectures and anomaly detection methods to preserve the safety and security of cyber-physical systems.
02 / Experience
A path from control
to applied intelligence.
Jun 2026 — PresentCurrent role
AI Researcher
Ericsson Canada · Montreal, Canada
Research in physical AI, telecom foundation models, integrated sensing and communication, and quantum-inspired optimization for complex AI and network problems.
Jun 2025 — Jun 2026
Postdoctoral Researcher
ÉTS & Ericsson Canada · Montreal, Canada
Developed quantum-inspired, gradient-free methods for modulation and coding scheme selection, and machine learning pipelines for error vector magnitude prediction.
Jan 2024 — May 2025
Machine Learning Research Intern
Ericsson Canada · Montreal, Canada
Developed parallelizable optimization strategies for deep neural networks and uncertainty-aware anomaly detection using conformal prediction.
Sep 2020 — Dec 2024
Control Systems & Cybersecurity Specialist
PreCySe Research Group · Montreal, Canada
Designed data-driven control architectures and PCA-based detection of coordinated cyberattacks. Applied variational autoencoders to naval-vessel anomaly and fault detection in a Canadian defence research project.
Apr 2017 — Aug 2021
Machine Learning Engineer
MAPNA Group · Karaj, Iran
Built LSTM models for wind-turbine predictive maintenance and deployed cloud-based anomaly detection combining SCADA and condition-monitoring data on MAPNA Mind.
Apr 2017 — Sep 2018
Control Systems Specialist
MAPNA Group · Karaj, Iran
Developed a DFIG wind-turbine simulator and hardware-in-the-loop simulations using MATLAB xPC Target and OpenFAST.
03 / Selected applications
Methods meet practice.
Telecommunications
Efficient learning for wireless systems
Quantum-inspired optimization for model training and MCS selection, alongside predictive models for wireless signal distortion.
Ericsson Canada · ÉTSCyber-physical security
Safety under network attacks
Data-driven safety and reference tracking for constrained systems, supported by robust control and anomaly detection.
Read the researchDefence & resilience
Anomaly detection for naval vessels
Variational-autoencoder-based detection of anomalies and faults, supporting early diagnosis in naval systems.
About the research program04 / Publications
Research in print.
Journal articles
- 2026
- 2024
- 2023
- 2023
- 2018
Preprint & conference proceedings 04
- 2019
- 2016
- 2016
- 2016
05 / Education
The foundations.
Ph.D. in Information & Systems Engineering
Concordia University · Montreal, Canada
M.Sc. in Electrical Engineering, Control Systems
Tarbiat Modares University · Tehran, Iran
B.Sc. in Electrical Engineering, Control Systems
Sharif University of Technology · Tehran, Iran
B.Sc. in Electrical Engineering, Control Systems
Hamedan University of Technology · Hamedan, Iran
06 / Contact
Good research starts
with a conversation.
For research collaborations and conversations about applied AI, optimization, and control systems.
mehran.attar@gmail.com