We can only accommodate a limited number of internships at any one moment. If you are interested to do an internship at VORtech, contact us to see whether there is a place available. Below is a list of internships. If there are no internships that match your interests, feel free to contact us anyway to discuss whether we can formulate a new project that is interesting both for you and for us.
Internship: Machine learning engineering
The field of machine learning is rapidly developing. As more applications become operational, things like machine learning engineering and MLOps become more important. This involves setting up a proper data pipeline, creating facilities to guard the validity of the model and to retrain it from time to time.
At VORtech, we try to keep up to speed with the new and updated tools in this field. Specifically, at this moment, we would like to try MLFlow and Airflow. The internship involves setting up a pipeline for a simple machine learning model with one of these tools (or other tools if you know better ones). We are particularly interested in the functionality for testing the robustness of the pipeline.
Internship: data assimilation
Data assimilation methods use sensor data to correct dynamic computer models to make them closely match the actual reality. This enhances their predictive power. A lot of the work that VORtech does in this field is based on the open source toolbox for data assimilation OpenDA. We have a number of ideas to further improve this toolbox and interns are welcome to contribute.
Apart from this general development, there is also work to do on one of the applications that was developed by VORtech: 4DCOOL. This is a digital twin of the indoor climate of data centers, used for optimizing the cooling. Data assimilation is the core of this application.
The internship concerns the improvement of the data assimilation in 4DCOOL. In particularly, we are looking for new methods that are both fast and robust for a wide range of data center topologies.
Internship: self learning calibration for traffic models using OpenDA
Traffic modelling is traditionally based on numerical models. In the past few years, a lot of interest went into data driven models: models that are exclusively based on data. But these data driven models cannot deal with unexpected or rare situations on the road. Therefore, there is an interest to still use numerical models but calibrate them using the available data. That would be the best of both worlds.
The traffic engineering consultancy firm Goudappel has an internship vacancy for automatic calibration of traffic models. One of the options is to do this with OpenDA, an open source toolbox that was specifically developed for this kind of application.
This internship for self learning calibration of traffic models with OpenDA will be supervised jointly by VORtech and Goudappel.