LightsOut

Inspiration

While brainstorming during a team call, a familiar household moment sparked our inspiration: one of our teammates was scolded by their parent for leaving lights on in empty rooms. This simple interaction made us realize that if managing energy waste is a challenge in a single household, it must be an enormous issue on a larger scale within our community. We immediately thought of our local schools and compared a parent enforcing household rules to a custodian managing a massive building. Instead of relying on manual enforcement, we envisioned a smart, data-driven approach. This led to the creation of LightsOut, a project centered around a 3D digital twin of Ardrey Kell High School. By inputting daily activity schedules, our system automatically optimizes lighting and climate control so that energy is only used when and where it is actually needed. Ultimately, LightsOut aims to eliminate this unnecessary waste, cutting our school's wasted electricity by thirty percent and saving upwards of one hundred and twenty thousand dollars annually.

What it does

LightsOut lets custodian and other operational staff enter clubs, meetings, exams, athletics, and events by room, time, and number of students. It shows a 3D school map where rooms are marked by energy and task status, then creates a custodian checklist for when to turn lights and HVAC on or off. It also estimates saved CO2 and recommends smaller available rooms when an event is scheduled in a space that is larger than necessary.

How we built it

We built LightsOut as a no-install static web app using index.html, styles.css, and app.js, so it can run on GitHub Pages. The 3D school model is made with HTML/CSS transforms. The local AI model trains in the browser using synthetic school energy scenarios and optional imported real kWh data. Weather-aware HVAC timing uses Open-Meteo forecast data.

Challenges we ran into

Our biggest hurdle began during a difficult ideation phase, as team disagreements over which project to build and conflicting schedules for setting up meetup times cost us massive amounts of limited hackathon time. Once we aligned on the idea, we ran into severe technical bottlenecks trying to integrate our APIs and AI models into the codebase, specifically encountering formatting and connection issues when deploying Gemini Flash alongside our backend logic, so we ultimately decided to ditch Gemini Flash.

Accomplishments that we're proud of

We are proud that LightsOut runs without installation, has a rotatable 3D school model, supports weather-aware HVAC timing, and includes a local AI model that can improve with real energy data. We also built a custodian-focused checklist instead of just showing raw analytics.

What we learned

Through building LightsOut, we learned that school energy optimization requires balancing a complex matrix of timing, weather data, and room size, such as moving a small club out of a massive auditorium to cut climate control loads. We discovered that AI is most effective when it supports human operations rather than replacing them, which we achieved by translating complex algorithms into a practical checklist for custodial workflows. Finally, because real-world energy logs were inaccessible, we mastered synthetic data engineering by building a pipeline to simulate energy curves and anomalies, allowing us to train our model directly in the browser.

What's next for LightsOut

Expanding LightsOut into a real-world school deployment involves connecting to smart meters, room calendars, and HVAC systems to enable automated temperature setbacks and collect actual kWh data. This continuous data streams into an automated pipeline to retrain the AI model over time, improving room-level accuracy and feeding a centralized admin dashboard that generates weekly reports on financial reductions and CO2 savings. Because campus schedules fluctuate, a bi-directional calendar sync and IoT motion sensors allow the system to adapt dynamically if an activity runs late, issuing proactive notifications to organizers asking if they want to extend services. Ultimately, a comprehensive notification system keeps teachers informed of upcoming shutdowns, alerts maintenance teams to abnormal energy draws, and engages the student body by broadcasting weekly sustainability milestones.

Built With

• css
• git
• github
• html
• javascript
• localstorage
• open-meteo-api