Company Information
Veridis is a deep-tech startup developing a novel instrument to improve quality control in plastic recycling: MADSCAN®. This MADSCAN® technology is a thermal analysis technique. This technology can be used to analyse plastics/polymers to give a material characterisation (material type, degradation, etc.). With our MADSCAN® technology we scale up the sample size from conventional technologies by a factor 100.000. This allows us to give a representative analysis of large batches of plastics, thereby increasing the relevance to the recycling industry. With this method we are able to analyse dark plastics, detect material degradation, and most importantly we directly increase the commercial value of recycled plastics.
Veridis has the ambitious goal to create a better future through science and technology. We are trying to create a place of work where scientists and engineers can develop technical opportunities for the benefit of society and climate. To achieve these objectives we are offering an internship with flexible working hours, an informal atmosphere, and a lot of independence and responsibility.
Assignment Description
The candidate will receive a fully implemented, though sub-optimal, Machine Learning pipeline implemented with the principles of Distillation from Weak Teacher (DWT) methodology. The candidate will be requested to do assisted and/or independent research on potential improvements for the pipeline with ultimate target the optimization of the systems’ performance in polymer fraction predictions. The project will include tasks such as model architecture optimization, fine-tuning, visualisation and interpretation of results. The code will be written in Python language and shall follow Software Engineering standards, giving the opportunity to the candidate to learn industrial-scale skills and collaboration best practices.
Intern Profile
You are a creative and curious person who is not afraid of a challenge and working with the inherent uncertainty that novel technological and software development brings. You like taking initiative and contributing new ideas about how to improve. Also, critical thinking and the ability to conceptually interpret thermal measurement data are required. You are able to develop software at an academic level and are capable of communicating the results, problems and next steps clearly.
We are very flexible in time requirements; we would like to fill the internship position full time, but part time is also a possibility.
Responsibilities
-
Research novel solutions that can improve the performance of the existing MADSCAN ML pipeline.
-
Write Python code to implement ML solutions and test them based on Software Engineering standards.
-
Document work in a clean way.
-
Collaborate using the version control system Git.
Requirements
-
Good knowledge of Python or other Object Oriented Programming language.
-
Good understanding of Machine Learning models such as NNs, RFs, SVMs.
-
Knowledge of Distillation from Weak Teacher (DWT) methodology.
-
Ability to conduct independent research of novel advancements in the field that can directly be applied and improve the MADSCAN ML pipeline.
-
Good communicational skills, team spirit and willingness to learn new things about Machine Learning, Computer Engineering and Data analysis.
Preferences
-
Knowledge/Understanding of Thermal Analysis.
-
Understanding of class and flow diagrams.
-
Having experience or understanding of Git, GitHub and version control tools.
Note that we can only consider students that are situated in the Netherlands during the duration of the internship. Please send your CV and list of grades to [email protected]
