Introducing Zeus!


In an ever-digitizing world, we continue to depend on electricity as a cornerstone for modern life. Global daily energy consumption is more than 1 million terajoules, putting a strain on the environment and on the people. This coupled with the most recent Texas blackouts following the February Winter Storms inspired us to create a forecasting companion to predict significant usage and costs in electricity across the major devices, utilities, and appliances in a home, building or office.

What it does

Zeus offers a variety of features to provide recommendations to organizations for conserving energy. The website’s primary feature is to use artificial intelligence to show the hours when power consumption is highest for the business, and what appliances consume the most energy. Once on the homepage of the website, the client has the option to either enter their own data for us to analyze, or to fill out a quick survey about their energy consumption habits. Based on this information, Zeus’s software will be able to generate a dashboard with reports that provides insights into energy usage trends of the client. From these reports, clients will receive recommendation plans that include forecasting for future energy usage, alternatives to reduce energy consumption, as well as, cost savings as a result of implementing the recommended alternatives.

How we built it


Our team spent the initial hour of the hackathon ideating and choosing which area of social good we wanted to impact. As a team, we are all very passionate about energy conservation and environmental sustainability. Thus, Zeus was created.


The Zeus website was created in React and hosted on AWS. The application of an AWS architecture was meant to offer means of scalability and resilience on a global level. This was done by hosting the web app on an EC2 instance and placing multiple copies in an Auto Scaling Group to increase/ decrease our number of instances automatically to match demand. This Auto Scaling Group was then placed inside an Elastic Load Balancer which would make sure that the traffic was balanced between each instance of our web app.


We pre-processed the data in Excel to make it ready for our AI/ML algorithms, Various machine learning, and artificial intelligence algorithms were tested in Python to analyze and interpret the data for this project. We extensively used Matplotlib, Pandas, and Turi Create to analyze the data. To connect our front-end to our back-end, we utilized Flask for its lightweight nature and speed.

Challenges we ran into

Finding datasets

The data we needed to find for this specific project was pretty difficult to find. Though we found several viable datasets, they were tangential to the specific goal we were trying to achieve. After extensive trial and error with multiple different datasets, we finally were able to first, find datasets with relevant data, and second, apply it to our programmatic goals.


For two team members, Niranjana and Sharan, this was their first hackathon. Though the last 24-hours were challenging, they learned so much from this great experience. Learning how to create an idea and develop its solution in such a short period of time was an invigorating and worthwhile experience.

Web Development

There were a lot of issues involved with testing environments especially following the addition of Flask onto our web app. Unfortunately, the web app’s front end remains dysfunctional but fixing this would be one of our biggest priorities going forward

Accomplishments that we're proud of

Along our journey, we were forced to step out of our comfort zone to combine different elements of building an AI software together. Some of us had some experience with the back-end elements, while some of us had a better understanding of front-end development. Being able to bring all these elements together to come up with an idea that helps with power conservation was a great way for us to understand the overall process of developing an AI application. This process also made us think about future plans and how we could improve our current product for more efficient use for clients. Some of the elements we took into consideration when discussing future plans included scalability and improving the user interface for easier navigation.

What we learned

One major learning curve for this project was understanding how to connect our front-end and back-end code. Though integrating these two entities was challenging, we were able to make it work after hours of brainstorming and trial and error. Additionally, because of the scope of this project, we were able to learn a lot more about energy conservation and how it impacts people and businesses. Energy runs the world and impacts every facet of our lives and creating this project helped us learn more about this field and how we can make a positive change.

What's next for Zeus

Scalability is the most important facet of AI applications, especially in those designed for social good. Currently, Zeus is targeted towards commercial buildings and the businesses they house. In the next 2 years, we hope to expand Zeus to all energy users especially those residential homes. Additionally, our next implementation goal is to take Zeus worldwide. Globally, energy consumption remains to be a leading cause of environmental harm and provides a significant financial burden on both institutions and people. Zeus will help people worldwide to better understand and interpret their energy consumption and will help the user make better energy choices.

Share this project:


posted an update

Our GitHub functioned primarily as a spot for experimentation for the front-end, back-end and a few machine learning models we theorized and would eventually implement in a coded solution/implementation of Zeus! While it isn't the primary solution, it shows our next steps if we were to move forward.

Log in or sign up for Devpost to join the conversation.