Internship Agentic AI for Automated Code Generation in Safety-Critical Embedded Software
Job Description
• During your internship, you will analyze our current development process (requirements, coding, unit tests, reviews and integration), as well as identify the steps with the highest automation potential.
• You will evaluate state-of-the-art agentic coding tools and frameworks (e.g. Claude Code, GitHub Copilot Agent Mode, MCP-based toolchains, multi-agent orchestration) for use with C/C++ on QNX, as well as embedded Linux.
• In addition, you will develop a process for creating an agent pipeline that takes a requirement or ticket and generates a pull request. This will include code, unit tests, as well as static analysis (MISRA C/C++, clang-tidy) and builds, as well as tests in CI and self-correction loops.
• Furthermore, you will familiarise the agents with our context using a structured knowledge base from the repository and architectural documentation, as well as coding guidelines, such as RAG, project rules and context engineering.
• You will work on the addressing safety and security constraints, such as ensuring traceability to requirements, conducting human-in-the-loop reviews, as well as ensuring compatibility with ISO 13849/IEC 61508 and CRA-driven processes.
• From the first day, you will be able to measure impact using clear KPIs, such as lead time per ticket, test coverage, defect rate, review findings and cost per feature. These will be compared against a manual baseline.
• Last but not least, you will deliver a working prototype, guidelines for the teams in Germany, Poland and India, as well as a recommendation for rollout.
Requirements
Function: Research
Experience Level: Internship