Vision Deep Learning Engineer – Pytorch

As a Vision Machine Learning / Software Engineer, you will play a key role in enhancing Mentium's deep learning library for vision AI models on our in-memory compute ultra-low power hardware. Collaborating closely with a small team of software and hardware experts, you will design and refine a client-facing codebase that enables seamless compilation and deployment of vision models for real-time applications. Additionally, you will support the training and testing of models for demonstrations. A strong understanding of vision model components is essential, including convolutional layers, transformers, activation functions, pooling, normalization, and non-maximum suppression. Familiarity with the structure, training and quantization of models such as YOLO, Mask R-CNN, MobileNet, ConvNeXt, EfficientNet, and Vision Transformers is required. You will work with libraries like Torchvision, Ultralytics, MMCV, Keras, ONNX, and OpenCV, adapting to client needs. Awareness of hardware fundamentals—such as memory transfer, inference time, model size, and precision—is critical for real-time edge deployments. Proficiency in Python is essential, including expertise in OOP principles, decorators, context managers, and numerical computing libraries like NumPy and pandas. You should demonstrate strong skills in modular development, collaborative coding using GitHub, and producing clear, well-structured documentation to enhance the customer experience. This role offers an exciting opportunity to work with cutting-edge vision AI technology and make a meaningful impact on the development of a robust, client-facing software library. Requirements and Qualifications: • Educational Background: Master’s or PhD degree in Computer Science, Electrical Engineering, or a related field. • Programming Expertise: Proficiency in Python and PyTorch, including object-oriented programming principles, decorators, context managers, and experience with numerical computing libraries such as NumPy and pandas. • Vision Model Proficiency: Strong understanding of components like convolutional layers, transformers, activation functions, pooling, normalization, and non-maximum suppression. • Model Experience: Familiarity with the structure, training, and quantization of models such as YOLO, Mask R-CNN, MobileNet, ConvNeXt, EfficientNet, and Vision Transformers. • Frameworks and Libraries: Experience with libraries like Torchvision, Ultralytics, MMCV, Keras, ONNX, and OpenCV. • Hardware Awareness: Understanding of hardware fundamentals, including memory transfer, inference time, model size, and precision, crucial for real-time edge deployments. • Development Practices: Proficiency in modular development, collaborative coding using GitHub, and producing clear, well-structured documentation. • Soft Skills: Strong problem-solving and analytical skills, excellent communication abilities, and a collaborative mindset to work effectively within a multidisciplinary team. • Familiarity with OpenCV programming and video processing pipelines is a plus. • Having a knowledge and experience on working with Camera drivers and camera pipelines using hardware accelerators such as ISPs is also a plus. Why Join Mentium? At Mentium, you'll be at the forefront of vision AI technology, working alongside a talented team dedicated to innovation. We offer a collaborative environment where your contributions directly influence the development of cutting-edge solutions. Join us to make a meaningful impact in the field of vision AI. Benefits: • Competitive compensation packages • Opportunity to work on diverse, cutting-edge AI projects across a range of industries. • 401(k) • Flexible PTO • Full PPO medical, dental, and vision insurance coverage Apply tot his job