Building Project Oracles for Hardware-Aware Software Development using Domain-Adaptive AI

/ Building Project Oracles for Hardware-Aware Software Development using Domain-Adaptive AI

Building Project Oracles for Hardware-Aware Software Development using Domain-Adaptive AI

Master internship - Leuven | More than two weeks ago

Shape the future of engineering productivity by creating AI systems that make deep hardware expertise instantly usable in software development

At imec, the SystemSoftware team builds applications that interact with complex hardware systems,from implantable neural probes to lens-free imaging platforms, DNA sequencers,and edge computing systems. Developing and maintaining this software requiresdeep knowledge of the underlying hardware: register maps, pin configurations,timing constraints, and countless other details scattered across datasheets,schematics, and specification documents.

The System Softwareteam works with a rich body of technical knowledge spanning hardwaredocumentation, design specifications, and implementation details. Unlockingthis knowledge effectively is key to faster development and better engineeringdecisions. While AI-powered coding assistants such as GitHub Copilot cansupport code generation, they often need deeper project-specific context to betruly effective in our domain. In this internship, you will help bridge thatgap by building "project oracles" that make project-specificknowledge more accessible and actionable throughout the development process.

During thisinternship, you will explore how large language models can be adapted to ourdomain, build a prototype that turns technical project documentation into apractical knowledge source, and evaluate its value for real engineering usecases.

You will also examinehow this prototype can fit into developers’ workflows and contributerecommendations for applying these techniques effectively within the team.

By theend of the internship, you will deliver a working oracle for one project, areproducible architecture that can be extended to other projects, and a set offindings and recommendations on the most effective techniques for our use case.

What wedo for you

imec is aworld-leading research and innovation hub in nanoelectronics and digitaltechnologies. In the Innovative Systems and Software Applications (ISSA)department, you will collaborate with experts in software engineering who buildthe tools and applications that bring imec's hardware innovations to life. Youwill work in the System Software team within the Software Development &Prototyping (SDP) group, where you will be embedded in a team of experienceddevelopers working on real-world systems across domains such as neuroscience,imaging, and genomics.

We offera stimulating environment at the intersection of AI and embedded systems,hands-on experience with state-of-the‑art LLM techniques, and the opportunityto build something that directly impacts how our engineers work every day.

Who youare

• You are a MSc student in ComputerScience, AI, or a related field
• You are available for a period of 3-6 months (share your available period in the application)
• You can communicate clearly and workindependently
• You have good written and verbalEnglish skills
• You enjoy coding in Python and areinterested in writing clean, reusable code
• You are curious about AI, largelanguage models, and how they can be applied to real-world engineeringproblems
• You are eager to learn, takeinitiative, and deliver tangible results
• Experience with RAG frameworks, LLMfine-tuning, or vector databases is a plus but not required
• Familiarity with embedded systems orhardware-adjacent software development is a bonus

Type of internship: Master internship

Duration: 3 Months

Required educational background: Computer Science, Electrotechnics/Electrical Engineering, Other

Imec allowance will be provided.

At imec, the SystemSoftware team builds applications that interact with complex hardware systems,from implantable neural probes to lens-free imaging platforms, DNA sequencers,and edge computing systems. Developing and maintaining this software requiresdeep knowledge of the underlying hardware: register maps, pin configurations,timing constraints, and countless other details scattered across datasheets,schematics, and specification documents.

The System Softwareteam works with a rich body of technical knowledge spanning hardwaredocumentation, design specifications, and implementation details. Unlockingthis knowledge effectively is key to faster development and better engineeringdecisions. While AI-powered coding assistants such as GitHub Copilot cansupport code generation, they often need deeper project-specific context to betruly effective in our domain. In this internship, you will help bridge thatgap by building "project oracles" that m