Om K

sr software engineer

kpit • Full-time

Aug 2021 - Feb 2026 • 4 yrs 6 mos

Developed and deployed real-time embedded software solutions on ARM-based microcontrollers and embedded Linux platforms, focusing on performance-critical and safety-oriented systems. Designed and implemented end-to-end Edge AI inference pipelines on NVIDIA Jetson and Raspberry Pi, including model conversion (ONNX/TensorFlow Lite) and runtime optimization. Optimized deep learning models using TensorRT (FP16/INT8) and quantization techniques to improve inference latency and reduce memory footprint on resource-constrained hardware. Built vibration-based predictive maintenance system using accelerometer data, implementing signal preprocessing (FFT, RMS, statistical features) and on-device ML classification for engine health monitoring. Engineered real-time data acquisition pipelines integrating sensors (SPI/I2C/UART) with structured buffering and latency-aware processing. Implemented camera-to-inference computer vision pipelines using OpenCV for object detection and real-time obstacle identification on edge devices. Performed performance benchmarking and profiling (CPU/GPU utilization, memory usage, thermal monitoring) to ensure stable long-duration edge AI deployments. Developed embedded communication stacks (CAN, LIN) and complex device drivers in AUTOSAR-based ECU environments with strict MISRA compliance. Designed modular system architectures separating acquisition, preprocessing, inference, and communication layers to enable scalable and maintainable deployments. Created automation and validation tools to improve regression testing efficiency and system reliability in embedded and AI-integrated environments.