Home/Case Studies
Selected Engineering Work

Engineering Case Studies

Real engineering outcomes across automotive validation, HIL automation, connected products, and intelligent embedded platform development.

Automotive · AUTOSAR

ECU Software Development & Validation

Full BSW stack configuration, MCAL driver development, and systematic validation for an automotive ECU programme targeting ASIL-B.

BSW stack configured on Aurix TC3xx
MCAL drivers: CAN-FD, SPI, ADC
UDS diagnostics integration and validation
ISO 26262-aligned test documentation
Classic AUTOSARMCALAurix TC3xx
Validation · HIL · Automation

HIL Test Automation for ECU Regression

Built and automated a complete HIL test bench with Python/CAPL scripting, Robot Framework, and nightly CI regression pipeline.

200+ test cases automated
Nightly CI with auto HTML reports
Fault injection and DTC validation
Integration with GitLab CI
HILCAPLRobot Framework
IoT · Connectivity

Industrial IoT Gateway Platform

Multi-protocol gateway with RS485, LoRa, cellular, MQTT cloud integration, and secure OTA — deployed in industrial monitoring.

RS485 + LoRa + Cellular on one device
MQTT cloud data pipeline
Signed OTA update system
Field-deployed in industrial environment
IoTLoRaESP32
Embedded Linux · BSP

Custom Yocto BSP for Automotive SoC

End-to-end Yocto BSP development for an NXP i.MX8 based automotive platform — from bare metal bring-up through secure boot and AGL integration.

Custom Yocto layer development
U-Boot secure boot chain
PREEMPT_RT kernel integration
AGL + VSS integration
YoctoNXP i.MX8AGL
Cybersecurity · HSM

Secure Boot & HSM Integration for ECU

Implemented measured secure boot chain and HSM-based key management for an automotive ECU programme targeting ISO 21434 compliance.

Full secure boot chain implemented
HSM key provisioning workflow
dm-verity rootfs integrity
ISO 21434 documentation support
CybersecurityHSMISO 21434
Edge AI · Inference

TensorRT Model Deployment on NVIDIA Orin

Optimised a YOLO-based object detection model for deployment on NVIDIA Orin using TensorRT INT8 quantisation and Triton Inference Server.

3.2× throughput vs FP32 baseline
INT8 PTQ with calibration dataset
Triton concurrent model serving
Real-time camera pipeline integration
TensorRTINT8NVIDIA Orin

Have a similar engineering challenge?

Talk to Noveltronix — we are happy to discuss your programme and share relevant experience.