Master Thesis – AI-Based Search for Reusable CAD Parts

Toyota Material Handling Europe; Linköpings kommun; On-site; SEK 179,000 - 246,000; Part time

Job summary

Toyota Material Handling Europe (TMHE) invites two Master’s students to join the DART research project in Linköping. You will explore CAD retrieval, geometry graphs and vision-language models to enable smarter reuse of components and improve engineering efficiency.

The role is based at the TMHE Innovation Centre in Vallastaden and the TMHE Headquarters in Mjölby, with close collaboration to Linköping University.

Qualifications

• Experience with Python is required.
• Ability to communicate results in English.
• Collaborate with engineers to understand practical needs.

Responsibilities

• Review CAD retrieval research, shape similarity and vision-language models.
• Create a benchmark of realistic search requests with input from engineering experts.
• Assess whether generated descriptions capture geometric features that matter for reuse.
• Compare retrieval methods based on structured geometry graphs, text embeddings, geometric shape descriptors and combinations of these approaches.
• Measure search quality, processing time and cost as the catalogue grows.
• Develop explanations of suggested matches, highlighting shared and differing features.
• Validate the approach with engineers at Toyota Material Handling.

Skills

Python

Education

Master's degree in Mechanical Engineering

Master's degree in Computer Science or Artificial Intelligence

Tools

CATIA

STEP files

Job description

At Toyota Material Handling Europe (TMHE), we create the technology that keeps the world moving. Diverse businesses across Europe depend on our logistics and material handling solutions. By pioneering technology such as automation, connectivity and productivity solutions, we create opportunities for our customers’ logistics operations to be as efficient as possible.

We firmly believe in setting ideas in motion. That is why we are investing more than ever in research and development. Our multidisciplinary teams work both conceptually and practically to develop the next generation of material handling products and solutions. By providing our Research and Development organisation with the creative freedom and resources needed to explore new technologies, we aim to stay at the forefront of innovation. At Toyota Material Handling Europe, we contribute to a smarter and more sustainable society, today and tomorrow.

Background and purpose

Finding existing CAD parts often depends on knowing a filename or part number. This can lead engineers to design new components even when a suitable one is already available, adding unnecessary work and cost across design, verification, purchasing and inventory.

Within the DART research project, TMHE and Linköping University have jointly developed a prototype that lets engineers search for parts by describing what they need. It combines CAD data, images and AI-generated descriptions of geometry to identify potential matches.

The prototype currently works on example parts. Its performance across a larger industrial library and its suitability for everyday engineering use remain to be established. This thesis will help address that gap and support more effective reuse of existing components.

Thesis Description

• Review research on CAD retrieval, shape similarity and vision-language models.
• Create a benchmark of realistic search requests, with relevant parts identified by engineering experts.
• Assess whether the generated descriptions capture the geometric features that matter for reuse and identify common errors.
• Compare retrieval methods based on structured geometry graphs, text embeddings, geometric shape descriptors and combinations of these approaches.
• Measure search quality, processing time and cost as the catalogue grows.
• Develop clearer explanations of suggested matches, highlighting shared and differing features so engineers can assess their relevance.
• Validate the approach with engineers at Toyota Material Handling.

The expected outcome is an evaluated search agent, a reusable benchmark and practical recommendations for future industrial use.

Your profile

• Mechanical Engineering
• Design and Product Development
• Computer Science or Artificial Intelligence
• A related engineering field

Experience with Python is required. You enjoy solving technical problems, testing ideas and evaluating results critically. You are comfortable collaborating with engineers to understand practical needs and can communicate your methods and findings clearly in English.

Beneficial Skills (not Mandatory)

• Familiarity with CAD modelling and 3D geometry, particularly CATIA or STEP files.
• Knowledge of or a strong interest in large language models, vision-language models and agent-based systems.
• Familiarity with information retrieval, semantic search or machine learning.
• Experience designing benchmarks or comparing algorithms through structured experiments.

Our offer

At Toyota Material Handling Europe, you will work on a real engineering challenge and explore how machine learning could support future product development decisions.

You will be based at our Innovation Centre in Vallastaden, Linköping and the TMHE Headquarters in Mjölby, where you will receive guidance from experienced professionals and have opportunities to collaborate with engineers and other Master's students.

We believe in a flexible work environment that allows you to balance both your professional and personal life. At TMHE, you will have the opportunity to build valuable skills, contribute to impactful projects and be part of a collaborative, diverse and welcoming workplace. You will also receive competitive compensation in two instalments: one when you start your work and the other upon successful completion of your thesis.

Time for you to make a MOVE!

Terms

• Application: As soon as possible, or at the latest by the 30th of October 2026.
• Start date: During January 2027.
• Scope: 30 hp.
• Number of students: Two.
• Location: TMHE Innovation Centre, Vallastaden, Linköping (Main). You must be able to attend in-person meetings at our offices in Mjölby when required.

#EUROPE