Garen
Arabkirlian
I work the full stack underneath automated-driving validation — vehicle-level V&V and functional safety on top, but also the embedded systems, electronics, and signal analysis that sit beneath it: PCB design, vibration analysis, firmware, and the C/C++ and Python that ties it together. Based in Munich.
Precision at the signal level.
I'm a Systems Verification & Validation Engineer at HOLON, where I've led vehicle-level verification for the development and road release of a Level 4 autonomous vehicle — the safety-critical layer that has to be right before anything drives itself.
That spans functional safety verification with full traceability from safety goals to vehicle-level tests, fault injection on steering and brake-by-wire systems via relay-matrix hardware I built myself, and actuator performance assessment worked out directly with suppliers like Dana, Continental, and Nexteer.
I also built the automation underneath it: an Azure DevOps CI/CD infrastructure for HIL and vehicle testing that's now standard across more than 30 verification engineers, internally and at supplier sites — plus the CANoe/CAPL/Python frameworks that keep it repeatable.
None of that started with testing, though — it started with electronics and embedded systems. I've designed and ordered custom PCBs, written C/C++ firmware for BLE/WiFi sensor nodes on PSOC6 and ESP32, and spent years doing accelerometer-based vibration analysis with MATLAB signal processing for predictive maintenance — the same instinct for reading a signal that now shows up in how I debug a CAN trace or a fault-injection test case.
How I got to the test cell.
Test systems, built to be trusted.
A sample of the verification infrastructure I've built and led at HOLON, supporting a Level 4 autonomous vehicle's road release program.
L4 Vehicle-Level Verification
Led whole-vehicle verification for a Level 4 autonomous driving road release program, with functional safety test cases carrying full bidirectional traceability from safety goals through system requirements to vehicle-level validation.
Fault Injection — Steering & Brake-by-Wire
Designed and built both the software interfaces and the relay-matrix hardware architecture for controlled fault simulation on steering and brake-by-wire systems, including SDIC and actuator-intervention verification.
Vehicle Dynamics Measurement
Configured, calibrated, and integrated OxTS RTK/GNSS and IMU measurement systems for high-precision vehicle dynamics data supporting steering performance and safety-goal validation.
Supplier-Integrated Validation
Worked directly with Dana, Continental, and Nexteer to analyse validation results and calibrate control systems, and with functional safety partners like FKA to validate safety concepts.
CI/CD for HIL & Vehicle Testing
Architected and deployed an Azure DevOps CI/CD infrastructure for automated HIL and vehicle testing — self-hosted agents, pipelines, reporting — standardized across more than 30 verification engineers, with mirrored environments at supplier sites.
ECU.Test Report Automation
A Python parser that reads ECU.Test XML reports to extract .blf trace filenames and auto-fills the Trace column across a full SDIC test-results workbook — cutting a manual step out of every test cycle.
Weekend builds.
Personal projects, public on GitHub — mostly embedded systems and computer vision, built for fun and to stay hands-on with hardware.
ComputerVisionProject
A weekend OpenCV pipeline — HSV-based object detection and tracking, built and calibrated across a 4K IMX415 and a Logitech C925e.
AIWatchWinder
An AI-driven watch winder concept for luxury watch collectors — small in scope now, built to expand.
ConveyorRobot
A desktop robotic sorting system with real-time conveyor tracking.
NVS_Fs
A flash-driver software component implementing a non-volatile storage filesystem on ESP32-IDF.
5-DOF Robotic Arm
MG996R servos on a PCA9685 driver, STM32/Arduino firmware, an Xbox-controller Python interface, and computer-vision ball detection with HSV masking and perspective mapping.
STM32 → ROS2 Bridge
A serial bridge connecting physical button presses on an STM32 Nucleo-G474RE to a ROS2 animation, running ROS2 Jazzy in Docker on Apple Silicon.
Seen in action.
Short clips from a few of the builds above — IoT sensor hardware, a gamepad-driven vehicle interface, and a computer-vision bike fit tool.
Predictive Maintenance Sensor Node
ESP32-based IoT sensor node built for predictive maintenance — capturing vibration and condition data ahead of failure. Grew out of my master's thesis work at IOSS, later deployed in industrial factories.
Xbox Controller Vehicle Interface
An Xbox-controller interface built to drive vehicle controllability directly from a gamepad, for test and demo use.
Bike Fit Analysis
A DIY Python tool using MediaPipe Pose and a side-on IMX415 camera to measure back, shoulder, elbow, and knee angles — markerless road-position analysis for dialing in my own bike fit.
Where the signal-reading started.
My engineering thesis at CNAM — the IoT vibration-analysis work that later shipped into real industrial factories, and the direct root of everything I do with signals today.
Système de Maintenance Prédictive IoT
A dissertation contributing to predictive maintenance by demonstrating the potential of IoT-based vibration analysis systems to improve equipment reliability and maintenance efficiency — covering the design, engineering, and industrial deployment of the system built during my time at IOSS LLC, which shipped into major industrial factories across Lebanon.
Let's talk shop.
Open to conversations about automated driving verification, functional safety, or the odd weekend robotics build.