Software Development Graduate Engineer

Back to careers

Description

About Ceva SDT

Ceva's Software Development Tools group builds compilers, tools, and embedded software for AI, Vision, Mobile Broadband, and NB-IoT processors. We work across the full stack — from LLVM-based toolchains to MLIR graph compilers — enabling customers to get the best out of Ceva silicon.

The Role

Join Ceva's software tools team and contribute to widely used open source based toolchains while building tools that run on real hardware across millions of devices. Our team embraces an AI-agentic development flow — using AI coding agents as part of everyday engineering — and we want engineers who are excited to work this way from day one.

Responsibilities

Contribute to Ceva's open source based toolchain targeting custom DSP and NPU architectures.

Analyze and optimize toolchain output for performance and correctness across Ceva processor families.

Apply AI-agentic development practices in day-to-day work — using AI coding agents to accelerate development.

Collaborate with architecture, hardware, and software teams.

Why Ceva SDT?

Start your career working on toolchain technology deployed in millions of AI devices. You’ll be mentored by experienced engineers, work with cutting-edge AI-agentic development tools, and grow fast in a team that values technical depth.

Requirements

• B.Sc. or M.Sc. in Computer Science, Electrical Engineering, or related field.
• Strong C++ skills.
• Proficiency in Python.
• Good understanding of algorithms, data structures, and software engineering principles.
• Analytical mindset with strong problem-solving skills.
• Good communication skills; independent and self-motivated.
  
  

• Experience with AI-agentic development — using AI coding agents (e.g. GitHub Copilot, Cursor, Claude Code) in software development workflows.
• Exposure to software development through coursework, personal projects, or internships.
• Knowledge of AI/ML frameworks or neural network models.
• Familiarity with embedded systems, DSP or NPU architectures.