Computer Vision Research Internship: Image to Sequence Modeling (e.g. Transformers)

Scandit is seeking a Computer Vision Research Intern in Zürich to advance machine learning methods for image-to-sequence modeling using architectures like Transformers. The internship focuses on research in deep learning for visual understanding, with a hybrid work model.

This research-focused internship supports long-term efforts to improve performance, robustness, and generalization in large-scale visual applications. The project centers on deep learning architectures for image-to-sequence modeling, including Transformers and attention mechanisms, contributing to cutting-edge computer vision research.

Responsibilities

• Design and test new machine learning architectures for structured visual data.
• Assess different modeling approaches, including encoder–decoder and hybrid Transformer models.
• Explore methods to enhance model robustness, generalization, and reasoning across multiple views.
• Conduct systematic experiments, ablation studies, and error analyses to validate research hypotheses.

Requirements

• MSc or PhD student in Computer Science, Machine Learning, AI, or a related field with a strong research orientation.
• Strong understanding of machine learning theory, neural networks, and computer vision.
• Proficiency in Python and deep learning frameworks such as PyTorch.
• Hands-on experience designing, training, and evaluating neural networks, including CNNs and Transformer-based models.
• Excellent analytical and problem-solving skills, with the ability to interpret results and refine approaches.
• Familiarity with research best practices, including reproducibility, controlled experiments, and ablation studies.

Nice to have

• Previous research experience in computer vision, pattern recognition, sequence modeling, or image-to-sequence architectures.
• Experience with training large-scale models or working with foundation-style architectures.
• Contributions to academic publications, preprints, or open-source machine learning projects.

What the company offers

• Certified as a 'Great Place to Work' in 10 countries.
• A skilled and supportive team in a dynamic environment focused on computer vision and advanced technologies.
• Regular events including hackathons, summer parties, and company outings.
• Office located in the city center of Zurich.

About the company

Scandit empowers businesses by automating workflows and delivering actionable insights. The company fosters an inclusive environment that values diverse backgrounds and experiences, empowering employees to contribute meaningfully.

• Inclusive workplace that welcomes all employees.
• Commitment to diversity and empowering employees through varied experiences.
• Focus on innovation and automation to help businesses across industries.