Ai testbed Compettion (3).png

Estonian government AI testbed competition invites everyone with expertise in data science and analysis to join an experimental testbed to find solutions to real-world problems using open data.

We look forward to welcoming a range of specialists to come and work with us and become part of one of the most technologically advanced government projects in Europe contributing to the development of AI Gov stack that benefits people across the globe. The AI testbed competition encompasses various works in the fields of data science, machine learning, and language technology, that open a way to contribute and become a part of developing the open-source AI Gov-stack. The AI testbed competition includes 5 different sub-competitions:

5.png

 

The Estonian government is developing a virtual assistant called Bürokratt, which will revolutionize how people interact with a government. Bürokratt has already been pretty smart but is still learning to detect multiple intents.  
This competition aims to create a model that can detect multiple intents from customer messages sent to the Estonian government.
An accurate model will help us provide the best service quality to all Bürokratt users.
Read more about the challenge and join the competition on Kaggle.

The Estonian museums' information system (MuIS) contains almost 4 million objects (musealia) from all over the world. Not all objects are labeled ideally, and manual labeling would take years. This, however, significantly hinders the re-use of musealia and its preservation. 
The goal of this competition is to predict the *`type`* of the museum object based on its descriptive variables. 
The accurate prediction model would allow both museum workers and ordinary users to find the objects more easily, enhance the national memory and improve its preservation.
Read more about the challenge and join the competition on Kaggle. 

9.png
12.png

Future career heavily depends on the skills you own but sometimes it is hard to predict the entire spectrum of expertise you need for one or another position. Matching and mapping the skills has become a struggle not only for the people who are looking for a job but for the HR specialist as well. The better the match is done the more benefit both parties will get!
The goal of this competition is to detect skills that correspond to the texts about the forestry/wood industry and metal/machine industry. 
The accurate skill detection model would result in big time saving for analysts, school principals, personnel specialists, career counselors, and the common person planning their learning and career path.
Read more about the challenge and join the competition on 
Kaggle. 

The image bank of the city of Tartu includes 134 000 images which are divided into folders by the general topics. The images are not systematically labeled which makes searching for the right image complicated. Thus, the re-use of existing data is significantly hindered for everyone.The goal of this competition is to label the images from the image bank of the city of Tartu. The accurate prediction model would help us to make the public images searchable, accessible and useable.Read more about the challenge and join the competition on Kaggle. 

14.png
3.png

Each year, the Estonian Health Board inspectors are measuring the quality of drinking water in more than a thousand water stations. 
Today, there is no predictive water quality model that would enable the Health Board to prioritize the tests or react proactively to the deterioration of the water quality.
The goal of this competition is to predict the water quality in Estonian water stations based on the government’s open data. 
The accurate prediction model would enable to increase the quality and accuracy of the work of water inspectors and ensure consistently high drinking water quality.
Read more about the challenge and join the competition on
 Kaggle. 

Participate alone or with your team and contribute to the development of AI Govstack, making Estonia to become the next generation AI-powered government.