SocialTechnologyLab

Ons werk

Wij maken praktische, geavanceerde AI-oplossingen die betekenisvolle verandering teweegbrengen. Door onze interdisciplinaire expertise en samenwerkingsgerichte aanpak werken we samen met organisaties in verschillende sectoren om complexe uitdagingen aan te gaan en duurzame, impactvolle innovaties te realiseren. Hieronder zijn wat voorbeelden van onze initiatieven.

Brillenbus

2024—now

education

social initiative

A mobile optician service working to prevent learning delays by addressing visual poverty early. The Brillenbus visits schools to screen students, helping to identify those who need vision care. The initiative aims to give every child a fair chance at education by ensuring they have access to proper eye care, regardless of their background or financial situation.

Brillenbus

DeBot

2024

LLM finetuning

data pipeline

PDF parsing

reproducibility

A project for the Dutch Ministry of Internal Affairs focused on political debate simulation using fine-tuned LLMs. Created a data pipeline to process over 50,000 PDF transcripts, extracting and structuring more than 2,000,000 spoken messages. Successfully fine-tuned Alpaca and Bloom models to generate realistic Dutch political debates.

DeBot

Digitheek

2024

prototyping

context-aware RAG

user interface

A groundbreaking digital literacy initiative developed in partnership with Dutch public libraries. The project's centerpiece is an innovative course environment featuring AI-powered personalized learning paths. The system includes an intelligent assistant that helps users define learning goals and create customized course trajectories. A unique feature is its multi-modal response capability, combining textual explanations with relevant visual aids from a curated image database, including screenshots and UI elements, making digital literacy education more accessible and engaging.

Digitheek

map interface

geodata

open datasets

Combined open datasets to identify surfaces beneath which you can find shelter. Set up a pipeline that can run these calculations automatically for any region in The Netherlands. Built an intuitive map-first user interface that shows you the best shelters near you.

AI finetuning

Deep Learning pipeline

advising

Developed a prototype based on Google's Neural Weather Simulation model for long-range weather prediction. Created a generalized data and training pipeline for NWS training, enabling efficient training and adaptation of various AI weather models. Provided strategic consultation on hardware and infrastructure choices to optimize costs and experimental capabilities.

Partijplein.nl

2023—now

semantic search

PDF parsing

automated translation

user interface

An innovative platform making Dutch election programs more accessible through AI technology. The site allows users to read party positions in simplified Dutch or various other languages, including Frisian, Arabic, and Polish. Using advanced AI-powered semantic search, the platform intelligently interprets user queries to find relevant political positions, even when exact terms don't match. This tool has been recognized by national media for its contribution to making politics more accessible and helping voters make more informed decisions without the bias often present in traditional voting guides.

map interface

geodata

open datasets

A pioneering web and native app solution that helps users find sun-soaked terraces throughout the Netherlands. By combining open datasets, we developed an innovative system to calculate sun exposure for outdoor seating areas at bars and restaurants. What started as an Amsterdam-focused initiative grew into a nationwide platform with over 3,500 user-contributed terraces. The project gained significant media attention, featuring on national television, radio, and in newspapers, demonstrating its value to both businesses and consumers.

Social Technology Lab logo

SocialTechnologyLab

© 2025 Social Technology Lab
KVK: 93608411