CV
You can download a PDF copy of my complete CV here.
Education
- Ph.D. in Electrical Engineering, Northeastern University, Boston, USA (Sep 2021 – Present)
- GPA: 3.81
- Thesis: Scalable Method for Stability and Robustness Verification of Systems with NN Feedback using Positivity Constraints.
- M.Sc. in Aerospace Engineering, Sharif University of Technology (Sep 2017 – June 2020)
- Control Courses GPA: 3.75
- Thesis: Coupled Attitude-Orbit Dynamics and Control of a Solar Sail via Center-of-Mass Manipulation.
- B.Sc. in Aerospace Engineering, Sharif University of Technology (Sep 2012 – June 2017)
- Thesis: Attitude Control of an Under-Actuated Satellite by Two Reaction Wheels.
Research & Impact
- Novel Neural Network Bounding: Designed global & local NN bounding (depth‑agnostic) methods, enabling \(\sim 10\times\) faster bound computation and tight, input‑aware sector envelopes for controllers.
- Scalable Verification Pipeline: Built a scalable robust stability verification pipeline for NN‑in‑the‑loop systems under positivity constraints, achieving up to \(\sim 10^4\times\) speedup over state-of-the-art baselines.
- Safe-by-Design NN Controllers: Designed neural network controllers for systems with time delays and model uncertainty: the architecture enforces the required stability sector on the controller’s gain at the origin by construction, and a sector-regularized imitation-learning loss pushes the trained controller to stay inside the sector.
- Sim-to-Real Gap: Validated learning-enabled control frameworks through extensive simulation and hardware-informed experiments, bridging the gap between theoretical guarantees and real-world robotic performance.
- Tool Integration: Implemented state‑of‑the‑art verification (AutoLiRPA, JAX‑Verify) and bridged them with classical control toolchains (MPC/robust analysis) for reproducible experiments using MATLAB/Python integration.
- Crowd Dynamics & Fluid Analogy: Modeled large-scale population motion as continuum flow dynamics and designed optimal egress control policies, highlighting transferable techniques for airflow and transport phenomena.
- Space Robotics Optimization: Derived coupled attitude–orbit dynamics of a solar sail via center-of-mass manipulation, including interplanetary path optimization using SNOPT.
Open-Source Software
- nnsector: Python and MATLAB library that computes global and local sector bounds of feedforward neural networks for absolute-stability analysis of neural feedback loops; supports PyTorch models.
- nnstab: Python and MATLAB library that certifies exponential stability of positive feedback loops with a neural network controller and finds the largest certified region of attraction, using Metzler and Hurwitz tests instead of semidefinite programming.
- More on the Software page.
Work Experience
- Graduate Research & Teaching Assistant, Northeastern University (Sep 2021 – Present)
- Supported autonomous experimentation environments (Quanser, Franka Emika) and reproducible computational pipelines.
- Research in control, estimation, and learning-enabled autonomy.
- Control Systems Engineer, Safatco (Dec 2019 – Oct 2020)
- Reverse engineered and developed electromagnetic servo valves for fuel control systems.
- performed SolidWorks modeling, precision measurement, and tolerance validation.
- R&D Engineer, Sharif Technology Services (Satellite-building Project) (Apr 2018 – Oct 2019)
- Built and configured Hardware-in-the-Loop (HIL) and Software-in-the-Loop (SIL) testbeds to support system integration.
- Developed digital models for satellite attitude and position estimation using EKF and UKF.
- Designed and executed sensor calibration procedures for magnetometers, gyroscopes, and magnetorquers.
- R&D Engineering Intern, Sharif Technology Services (Satellite Reaction Wheel Project) (Jan 2017 – May 2017)
- Conducted R&D for a satellite reaction wheel; drafted verification and calibration procedures.
Selected Projects
Robotics, AI & Control
Autonomous Lab Deployment: Installed and calibrated multi-camera sensing systems in the Quanser autonomous lab, enabling real-time 6-DoF pose estimation for aerial and ground robots and supporting hardware-in-the-loop autonomy experiments.
Sensor Fusion & Control Integration: Integrated perception, EKF/UKF state estimation, and feedback control pipelines in a robotics testbed to enable reliable autonomous flight and driving experiments.
Quadrotor Real-Time Control: Designed and deployed controller–observer architectures for quadrotor dynamics on Arduino hardware, achieving real-time stabilization under noisy sensing conditions.
Ground-Effect Aerodynamic Modeling: Conducted CFD-based analysis of quadrotor rotor–ground interaction using Fluent/Gambit to inform control performance limits with physics-based insights.
Dual-Arm Manipulation: Configured and coordinated two Franka Emika robotic arms for interactive manipulation tasks (Tic-Tac-Toe), demonstrating synchronized multi-robot control.
Safe Imitation Learning: Implemented imitation learning of an MPC controller for nonlinear inverted pendulum dynamics, verifying closed-loop stability before deployment.
Safe Multi-Agent Control: Designed learning-based controllers for a network of quadrotors using a scalable NN-in-the-loop verification framework to provide formal stability guarantees.
Learning-Enabled Robotics: Developed control frameworks integrating neural-network feedback with model-based dynamics to enable trustworthy deployment of learned robotic policies.
Certified Autonomous Driving Control: Formulated vehicle lateral control under neural-network feedback using IQC-based parametrization to achieve formally certified performance bounds.
Neural Network Verification & Learning-Enabled Systems
Neural Network Sector Bounding: Developed global and local sector-bounding methods for feedforward neural networks, enabling depth-agnostic and input-aware certification tighter and faster than prior approaches.
Scalable Stability Verification: Built a positivity-based stability and robustness verification pipeline for NN-controlled dynamical systems, achieving up to \(10^4\times\) computational speedup over state-of-the-art.
Exact MILP Certification: Formulated mixed-integer (MILP) encodings of ReLU networks for exact closed-loop stability verification of nonlinear feedback systems.
Relaxation Analysis for NN Verification: Designed continuous relaxations of mixed-integer NN formulations to analyze tractability–accuracy trade-offs in verification.
Reachability & ROA Analysis: Performed reachability and region-of-attraction analysis of nonlinear systems with NN feedback using IQC-based formulations to provide formal safety guarantees.
Cross-Platform Verification Pipelines: Integrated AutoLiRPA and JAX-Verify with Lyapunov and MPC analysis via MATLAB–Python pipelines for reproducible certification workflows.
Time-Delay Neural Stability: Verified uncertain retarded neural-network feedback loops using positive Lur’e frameworks to extend stability certification to time-delay systems.
System-of-Systems Robustness: Developed verification methods for AI-enabled systems-of-systems under time delays and parametric uncertainty to strengthen safety guarantees in autonomous settings.
Spacecraft, Orbital Mechanics & GNC
Solar Sail Trajectory Control: Derived and controlled coupled attitude–orbit dynamics of a solar sail via center-of-mass manipulation using constrained nonlinear optimization (SNOPT) for interplanetary trajectory design.
Under-Actuated Satellite Control: Designed nonlinear attitude control for a satellite with only two reaction wheels, achieving stability under actuation constraints.
Fuel Sloshing Stability Analysis: Modeled and controlled satellite attitude considering internal fuel sloshing effects to improve robustness of stability analysis.
Nonlinear Spacecraft Momentum Control: Implemented feedback linearization for nonlinear 6-DOF spacecraft attitude dynamics to enable precise momentum-transfer control.
High-Fidelity GNC Simulation: Built integrated 6-DOF spacecraft GNC simulations with EKF/UKF and disturbance modeling for mission-level validation.
Orbital Mission Modeling: Developed ground-track, eclipse, visibility, repeated-track, and CR3B simulation tools for comprehensive spacecraft mission analysis.
Satellite Estimation Systems: Developed EKF/UKF-based attitude and position estimation pipelines for integrated spacecraft simulations and testbeds.
Reaction Wheel Engineering: Engineered verification and calibration procedures for satellite reaction-wheel hardware during aerospace manufacturing.
Aircraft & Fluid Dynamics
Aircraft 6-DOF Modeling: Developed nonlinear aircraft guidance and control simulations in MATLAB/Simulink for robustness analysis under disturbances.
Turbulence Robustness Analysis: Performed statistical analysis of aircraft encountering clear-air turbulence with measurement noise modeling to quantify control degradation.
Wing-Box Structural Simulation: Built finite-element models of aircraft wing-box structures using Abaqus/ANSYS for stress and modal analysis.
Re-Entry Aerodynamic Simulation: Conducted CFD simulations of re-entry vehicle aerodynamics using Fluent/Gambit to characterize high-speed flow behavior.
Thermofluid System Modeling: Modeled internal fluid flow systems using conservation laws to analyze pressure drop and heat transfer regimes numerically.
Networked & Data-Driven Dynamical Systems
Sparse System Identification: Applied SINDy and Sum-of-Squares methods to discover governing nonlinear equations from data and certify stability properties.
Ocean Dynamics Identification: Developed data-driven pipelines to model oceanic transport dynamics from drifter measurements as spatiotemporal systems.
Networked Epidemic Control: Designed centralized and decentralized vaccination strategies for SIS epidemic models using graph-theoretic optimization.
Adaptive Networked Control: Integrated adaptive identification with safe control for distributed dynamical systems to improve scalability of autonomy frameworks.
Distributed Estimation: Developed distributed adaptive estimation algorithms for sensor networks with partially unknown nonlinear dynamics.
Crowd Flow & Egress Optimization: Modeled crowd dynamics via fluid analogies and designed optimal evacuation strategies under uncertainty.
Mechanical & Industrial Systems
Gas Turbine Servo-Valve Redesign: Reverse-engineered and redesigned an electromagnetic servo-valve for industrial fuel control systems.
Fuel-Nozzle Test Rig Development: Designed and built a combustion-component test rig to improve experimental validation capability.
Industrial Bearing Modeling: Modeled and validated industrial bearing geometries in SolidWorks with tolerance analysis for manufacturing precision.
Computational & Systems Programming
Autonomous Game Engine: Implemented an automatic Reversi (Othello) game in C with rule-based decision logic and structured state representation.
Kalman Filtering Frameworks: Implemented multiple Kalman-filter-based estimation frameworks for nonlinear aerospace systems to improve robustness under noisy sensing.
Technical Skills
- Robotics & Perception: 6-DoF rigid-body dynamics, robotic arms, real-time control, multi-camera sensing, sensor calibration, EKF/UKF.
- Machine Learning: Imitation Learning, Reinforcement Learning, PyTorch, TensorFlow, AutoLiRPA, JAX-Verify.
- Control & Optimization: Nonlinear Control, MPC, Robust Control, Reachability Analysis, Lyapunov Methods, SNOPT, CVX.
- Programming: Python, MATLAB, C, Julia.
- Simulation Tools: Simulink, Fluent, Gambit, SolidWorks, Abaqus, ANSYS.
Teaching & Mentorship
- Teaching Assistant:
- Embedded Design Enabling Robotics (Spring 2026)
- Linear Systems Analysis (Fall 2025)
- Electrical Engineering Class & Lab (Spring 2025)
- Classical Control (Spring 2023)
- Mentorship: Capstone project “Robotic Gloves for Teaching Piano” (Summer 2024).
Professional Service
- Leadership: Founder & President of Northeastern ECE Ph.D. Council.
- Reviewer: IEEE CDC, ACC, Control Systems Letters, IFAC CPHS.
- Memberships: IEEE, Society for Risk Analysis (SRA).
