NOLAI tech blog

NOLAI is the National Education Lab AI. Together with schools throughout the Netherlands, businesses and researchers, we work on responsible educational AI for primary, secondary and special needs education. On this blog, we focus on the tech behind it.

Usually, we put education first: teachers and students are central, not the technology. On this blog the roles are reversed and the tech takes the stage. We write about technology in general, usually AI but certainly not only, about what we experiment with, and about what we learn along the way. We share our results as openly as possible.

This blog is for developers, tinkerers and tech enthusiasts of every kind: for anyone who is fascinated by technology and wants to develop and apply it responsibly, to solve real problems in society.

Join our Discord Check out our GitHub About NOLAI

15 min
Privacy Cryptography
Data sharing in Dutch education using the ECK iD
Data sharing in Dutch education using the ECK iD Nearly every pupil in Dutch primary education (PO), secondary education (VO) and vocational education (MBO) has an identifier that schools, publishers, distributors and learning platforms use instead of their name or citizen service number (BSN). It is called the ECK iD: a pseudonym of the BSN, generated by a central service called the Nummervoorziening. It exists so that schools and their suppliers can reliably share data about a pupil without passing around the BSN. This post explains why we have the ECK iD, how it is generated, who is …
Job Doesburg Job Doesburg
read more ▸
4 min
ROCm framework
Deploy your PyTorch app on the Framework desktop with ROCm in a Docker container
Deploy your PyTorch app on the Framework desktop with ROCm in a Docker container We, the NOLAI Tech Team, have bought 4 framework desktops. They are great since they have such a large amount of (shared) RAM and a powerful AMD GPU. But that same AMD GPU is also a bit of a headache when it comes to running PyTorch models. The CUDA ecosystem is so dominant that most PyTorch images and packages assume you have an NVIDIA GPU. But luckily AMD has a solution: AMD gives you a Docker image with ROCm and a version of PyTorch that should work as a drop-in replacement for the CUDA build. The reality is …
Julian van der Horst Julian van der Horst
read more ▸