Embedded AI Engineer
New York, NY
June 2023 — Present
Edge Computer Vision for Aerospace Inspection

• Developed and deployed edge computer vision pipelines for aerospace component inspection, including surface defect detection, connector alignment verification, and dimensional tolerance validation for metallic components, machined assemblies, and avionics connector systems using Python, C++, OpenCV, PyTorch, and ONNX/TensorRT inference pipelines.

• Deployed and optimized CNN/YOLO-based computer vision models on NVIDIA Jetson Xavier NX and ARM64 embedded Linux environments, reducing inference latency by ~30–40% through TensorRT optimization, asynchronous inference execution, and runtime profiling.

• Applied embedded inference optimization techniques including FP16/INT8 quantization, ONNX Runtime acceleration, and memory pipeline optimization to maintain stable detection performance under constrained edge hardware resources.

• Performed CPU/GPU profiling and runtime analysis using Nsight Systems and tegrastats to identify bottlenecks, improve thermal stability, and maintain more deterministic inference behavior on embedded devices.

Flight Data Analytics & Predictive Monitoring

• Built Python-based telemetry processing pipelines for vibration, temperature, and pressure sensor streams, supporting anomaly monitoring and operational diagnostics for aerospace engineering workflows.

• Implemented signal preprocessing, statistical thresholding, and trend analysis methods to assist engineering teams in identifying abnormal telemetry patterns and supporting early-stage fault investigations.

• Supported telemetry validation workflows integrated with MATLAB/Simulink-based engineering environments, improving engineering verification efficiency and reducing manual analysis effort.

Embedded Systems & Real-Time Constraints

• Integrated AI inference workflows into embedded Linux (ARM64) environments using containerized deployment and low-latency communication through UART and CAN interfaces.

• Implemented lightweight sensor synchronization and diagnostics pipelines to support reliable operation in constrained embedded environments with near real-time inference feedback.

• Monitored and optimized CPU/GPU utilization, thermal behavior, and memory constraints to maintain stable inference operation during prolonged edge deployments.

Safety & Certification Support

• Supported software lifecycle, verification, and documentation activities in DO-178C DAL-C aligned development environments, contributing to requirements traceability, software testing evidence, and engineering documentation.

• Worked with IBM DOORS requirements workflows and participated in engineering activities aligned with ARP4754A development processes.

• Supported certification-related activities for surrounding software infrastructure and validation environments, while ML inference components were utilized as non-certified assistive systems.

Key Achievements

• Reduced embedded inference latency by ~30–40% on Jetson Xavier NX through model optimization and deployment tuning.

• Improved automated inspection consistency and reduced manual quality review effort through deployment of embedded computer vision workflows.

• Increased runtime stability during prolonged edge operation by optimizing memory utilization, thermal behavior, and inference execution pipelines.

Key Skills
Python, C++, OpenCV, PyTorch, TensorRT, ONNX Runtime, Embedded Linux, NVIDIA Jetson, ARM64, Computer Vision, Edge AI, Docker, UART, CAN, MATLAB/Simulink, DO-178C, DAL-C, ARP4754A, IBM DOORS, Real-Time Systems, Nsight Systems