Job description

At DAF we use model-based engineering to develop software for our vehicle and powertrain controller. Throughout this process, we collaborate with colleagues from various departments who provide us with high‑level requirements. These requirements describe desired functionalities and system behavior together with their allocation into different software components.
As part of our current workflow, these high-level requirements are translated into detailed, low-level requirement documents. These low‑level requirements describe down to the exact logic, the expected implementation in our software models. Creating these documents, is a manual and time-consuming process.
The assignment should bridge the gap between our current workflow and a future in which we write down proper requirements. The goal is to find out if we can automate parts of the described workflow using a large language model (LLM) based tool. The tool should be able to produce consistent, correctly styled low-level requirement suggestions, reducing engineers workload.
The assignment should also address one of the risks of having assisted requirement writing. Engineers may rely on the automated suggestions, without critically reviewing the suggestions. Your solution should therefore explore strategies that encourages the user to critically review each suggested requirement.
This task includes:

• Analyzing examples of existing low-level requirements to extract structural, linguistic, and stylistic patterns
• Researching and evaluating the feasibility of using LLMs to produce low-level requirements
• Implementing a prototype tool that, based on high-level requirements, suggests low-level requirements in correct format and style.
• Addressing the risks of having engineers being overly reliant on the automated suggestions.

• Automatically uploading the generated low-level requirement into our requirement management system, including trace linking to the relevant high-level requirement.