Empowering Local Languages

AI Challenge

This challenge invites participants to design and develop advanced Artificial Intelligence models capable of translating an Indigenous language from the northern regions of Australia into English. It is more than just a technical competition; it is an opportunity to bridge cultures, preserve linguistic heritage, and make Indigenous voices more accessible to the world.

Registration closes on 25 Sep 2025

The challenge mirrors a complete AI project lifecycle, from data preparation and model training to deployment-ready evaluation, with leaderboard results updated regularly. Participants will gain hands-on experience in natural language processing, low-resource language translation, and culturally aware AI applications, all of which are valuable for both research and industry careers.

Participants will explore a variety of machine learning and deep learning techniques, including sequence-to-sequence models, transformers, and attention-based architectures, to produce high-quality translations. Creativity in model design, data preprocessing, and handling linguistic nuances will be essential to achieving the best results.

By developing an AI model capable of translating the Kriol language into English, participants will not only showcase their technical skills but also play an active role in a cultural preservation initiative.

Participants will train and validate their translation models using the provided training dataset (accessible via a shared link), while the test dataset will remain private for final scoring and ranking by the IT Code Fair team.

You will showcase your project to industry judges in person on the Challenge Day date and venue as stated below.

What is Required?

Participants must use the training dataset available at <link> and may combine it with any publicly available datasets or pretrained weights to improve model performance.

Rules of Competition

Participants must be in teams of 2 to 4 students.

The solution must be developed in Python using the PyTorch framework. The trained model must be submitted as a .pth file along with the Python source code.

The submission code files in .ipynb format is accepted. Use of Google Colab or any interactive Python editor is recommended

Submission

Please submit your substantial work-in-progress by the Project Submission Deadlines stated below. You may include Python scripts, all processed datasets, and the requirements.txt file (if applicable) to replicate your Python environment.