• Engineered real-time video analytics pipelines using NVIDIA DeepStream SDK, integrating ONNX object detection models to process road traffic data with low latency.

• Leveraged hardware acceleration on NVIDIA Jetson platforms to execute inference, utilizing CUDA-backed libraries within GStreamer plugins to maximize throughput and frame rates.

• Architected a decoupled microservices system using Docker, separating the C++ inference engine from cloud connectivity to ensure non-blocking operation and system stability.

• Optimized embedded C/C++ applications for resource-constrained Linux environments, refactoring code to minimize memory footprint and prevent race conditions in multi-threaded processing.

• Developed hardware-level firmware for STM32 microcontrollers, writing efficient drivers for peripherals (ADC, DAC, GPIO) and communication buses (I2C, SPI, UART) without relying on heavy abstractions.

• Refined the build and deployment workflow using AWS IoT Greengrass and Docker, enabling reliable Over-the-Air (OTA) updates for a fleet of edge devices.