Computer Vision Engineer

The role of a computer vision engineer involves analyzing RGB and 3D camera streams to identify human key markers such as parental interventions, baby presence, baby sleep states, and health analytics within a resource constrained embedded system. This is achieved by utilizing a combination of classical DSP and neural networks to translate scientific concepts like optical flow, signal extraction, and real-time inference into code that runs on embedded Linux. One of the responsibilities includes leading data collection campaigns by setting up annotated video datasets under various lighting conditions. Analyzing vast amounts of vision data for patterns in both 3D and RGB formats is also a crucial aspect of the role. Additionally, building and refining computer vision pipelines in Python using OpenCV or neural networks to detect and track baby movement, posture, and breathing is essential. Furthermore, the role requires cross-compiling and deploying vision modules to ARM-based SoCs while optimizing for CPU/GPU constraints and power budgets. It would be advantageous for candidates to have 5-8 years of prior experience in computer vision engineering and signal processing. Proficiency in Python and OpenCV for tasks such as image processing, filtering, optical flows, SVD, and PCA is highly desirable. Hands-on experience with deep learning vision frameworks like TensorFlow and PyTorch, as well as model quantization for edge inference, is considered a plus. A strong understanding of traditional DSP techniques and linear algebra is also beneficial for this role.,