About the Business Unit:

The Vision Business Unit (VBU) is a unique department in Ceva combining a dynamic startup working environment with a stable and well positioned IP company. A VBU Software organization goal is to create and supply the most cutting-edge embedded software and graph compiler core for AI and Vision applications. We are a forward-thinking, innovative organization that continuously pushes technological limits to enable widespread access to fields like augmented reality, virtual reality, 3D processing, and Generative AI.

About the Role:

In this role, you will learn and deeply understand the AI domain including neural network models, quantization techniques, network performance and precision. You will develop most advanced Neural Network Graph Compiler tools that will be used by CEVA customers across the globe.

Responsibilities:

Develop and implement software compiler for deployment of AI models on proprietary hardware. Optimize high-level algorithm implementations for CEVA NPU and DSP architecture. Independently manage projects from start to finish, ensuring timely delivery. Collaborate with cross-functional teams to integrate solutions into products. Conduct thorough testing and debugging of software. Stay updated with the latest advancements in AI compiler technologies.

Requirements:

B.Sc or M.Sc in Engineering, Computer Science, or related technical field.
At least 3 years of proven experience as a Python developer and knowledge of C++
Excellent problem-solving skills and creativity in software design
Ability to work independently and manage dynamic environment with interrupts
Strong organizational skills and attention to detail
Effective communication skills and ability to collaborate with team members.

Advantages:

Experience in Graph Compiler engineering
Experience in embedded systems development
Familiarity with IR, optimization, and quantization
Strong knowledge of CI/CD, git, pytest, dockers, CMake and torch
Programming in Linux environment
Knowledge in AI modeling and neural networks.