The Super Models

Inspiration

Our team at servian is passionate about solving problems and were excited to apply our problem solving skills for good. Our team was inspired by our leaders and the wider community to not only do good for our industry but the wider world and to help clients, customers and the community to use technology to do inspiring and good things.

In order to uncover our key direction we took a collaborative approach where we went through a discovery phase. During this discovery process we went through an iterative idea generation process in order to narrow our scope. Collaborating through a virtual whiteboard space we narrowed down our direction to combat climate change and finding a way to increase the uptake of renewable energy.

Climate change and energy usage is of particular interest to us and we believe that technology, in the form of Machine Learning models, can help the public, government and industry make better decisions about what energy to use in which locations to provide the best cost to energy usage. That industry and governments can use this information to smartly incentivise and invest in the right renewable energies in the right locations.

What it does

Our Application has been developed to provide an easy to use map that anyone can request a prediction of how much energy could be produced from solar energy at a given location. The backend services have been created to enhance the model with wind, water, coal or eleven hydrogen generation predictions in a way that will allow comparison based predictions to enhance decision making.

By providing easy access to this information around the optimal renewable energy to use in locations will lead to clarity and an increase in the uptake of renewables.

The MLOps pipeline has been built to develop, deploy, maintain and monitor the Models to provide these predictions in a way to allow rapid iteration while providing good efficacy of predictions

How we built it

As we worked to collate the relevant data required to build our solution we also moved into a rapid prototyping phase to help to inform and functionality of the user experience and the user interface design.

We used an existing model and solar data in New Zealand for the last 2 years. We then instantiated a MLRun instance in the provided Azure environment, also deploying Application containers to house the Front End Map.

Within the MLRun environment we configured a Gitlab repo backed MLOps pipeline

Although not all of the orchestration and automation features were implemented the architecture would be advanced over time to provide the functions.

Challenges we ran into

Our biggest challenges fell into 2 categories: Time and resources Models and data Time and resources, although the team dedicated many hours Remote ways of working

In order to maintain our momentum and meeting the project deadline we decided to narrow to scope of your project to the locations of Australian and New Zealand and the renewable However, the future state intention of the platform is to include all renewables and the world to allow everyone easy access to this information to encourage the uptake of renewables across the globe.

Accomplishments that we're proud of

Collaborating as a team utilizes virtual white space and video calls allowing us to work across varying locations.

What we learned

The importance of collaborating with team members with varying skill-set and backgrounds to have a more holistic approach

What's next for The Super Models

We are passionate about bringing all forms of available renewable energy into the platform. We also intend to scale it to other locations, allowing the information to be easily accessible to a number of countries.