Overview Project description
Our client, a leading manufacturer of high-end household appliances, is expanding into new smart product lines. As part of this strategic initiative, a large-scale global program is being implemented across the company's IP portfolio. This includes the development of new embedded software, enhancements to cloud infrastructure.
Responsibilities Work directly with computer vision perception engineers, and camera ISP engineers.
Define the image quality requirements for customer use cases.
Develop image processing algorithms to optimize the camera image quality for performance improvement of computer vision applications.
Develop custom ML models for visual recognition and training them from scratch.
Quantization and pruning the ML models.
Data evaluation using relevant metrics to increase the model performance.
Implement computer vision/image processing algorithms in fast/optimal C++ code.
Implement heuristic/math algorithms and probabilistic methods for camera AI.
Research existing C++ solutions for CV logic.
Develop cross platform C++ code.
Cover algorithms and CV code with unit tests.
Ramp up on new technologies related to CV/ML/algorithms (not R&D).
Integrate machine learning modules into algorithms.
Skills Must have
Bachelor's Degree or greater in Computer Science, Software Engineering, or related field.
3-5 years of experience developing computer vision applications/algorithms.
Participate in at least 2 image processing / computer vision projects and 1 of them productized.
Image processing / computer vision background/understanding.
Strong C/C++ experience.
Experience with OpenCV and ML frameworks such as PyTorch, TensorFlow, TensorFlow Lite.
Deep understanding of neural network architectures.
Experience to work with embedded devices in C++ on Linux.
Decent understanding of geometry 2D/3D and algebra.
Desire to learn new technologies.
Desire to follow test driven development (in CV/ML too).
Nice to have
Experience with machine learning integration, inference engines.
Machine learning general understanding.
Building for Android/Yocto experience.
OpenCL/CUDA or OpenGL/Vulkan (shaders) experience/understanding and desire to learn or any different SIMD language.
#J-18808-Ljbffr