Statistics Iceland AI Data Hackathon – Building an Intelligent Statistical Assistant

The objective of the hackathon is to develop a working prototype of an AI-powered statistical assistant that allows users to ask questions about Statistics Iceland’s data in natural language and receive clear, evidence-based answers. The solution should combine a large language model with RAG, direct access to Statistics Iceland’s PX-Web API, or both, together with a user-friendly interface such as a website chatbot.

Participants may run their backend on an NVIDIA DGX Spark, a powerful GPU-equipped computer, or the LUMI supercomputer. Solutions may use a local vector database, retrieve the latest figures directly from Statistics Iceland through API queries, or combine both approaches. Necessary data and metadata shall be extracted from Statistics Iceland open database, while suitable computing resources will be made available through University of Iceland and the Kveikja and Glóð projects.

The hackathon will run for 48 hours and is open to university students, professionals and companies. Participants should have solid general programming skills and experience developing systems from the ground up. Familiarity with LLMs, LoRA, RAG, AI agents and tools such as Python, PyTorch, TensorFlow, Hugging Face, Ollama and GitHub is beneficial, but advanced expertise in every area is not required.

Expected outcome: A working prototype that receives a user’s question, identifies the correct statistical table and relevant variables, retrieves the required figures, and produces a clear natural-language response with a transparent reference to the official statistics published by Statistics Iceland.

View an example of the Hag statistical chatbot.

Is this hackathon for you?

You should be a confident programmer who can contribute to the design and development of an end-to-end AI solution. Experience with Python and one or more of the following is desirable: LLMs and LoRA, RAG, vector databases, APIs, PyTorch, TensorFlow, Hugging Face, Ollama or similar tools.

Individuals and teams are welcome to register.