Back on Track

Main Page
Tutorial
Upgrades
Gameplay

Inspiration

We were inspired by factory games like Factorio and Mindustry which we felt were close to the Industrial Revolution with a mix of Mini Motorways gameplay - a favorite.

What it does

Game meant to promote problem solving and resource management in prospective engineers. The game requires creativity and planning ahead as you turn a bare map into a working steampunk industrial network.

We integrated the Gemini API to make gameplay more dynamic and personalized by analyzing player statistics (e.g. progress, finances, completed orders, and infrastructure) to generate adaptive orders and upgrades that adjust to the player's skill level and performance. It also helps create end-of-game reports that reflect players' experience, while existing hardcoded content serves as a fallback if the API is unavailable.

We also integrated ElevenLabs text-to-speech to read tutorials, pop-ups, instructions, and other in-game information aloud. Players can toggle narration on or off through the settings menu, alongside controls for gameplay audio and volume.

How we built it

Back on Track is a web application build on React + Tailwind with Vite and TanStack Start Lovable.ai was used to setup the tech stack and map generation was made with the help of Open Street Maps with the Overpass API to grab real data from Long Island. The 2.5D visualization was made possible with Open-Meteo elevation API and Shapely to round out and improve geometry into a more user-friendly design. All map data is saved to a JSON to prevent callbacks to APIs during runtime.

Challenges we ran into

We had initial trouble brainstorming and starting the project but we eventually managed to come to a consensus and stat on this wonderful project!

Accomplishments that we're proud of

All art used in the project was made by humans! Every sprite, tile, and UI element was drawn by our team, which gave the game a cohesive steampunk look with brass, soot, and glowing boilers that feels like it belongs to one world.

Beyond the art, we’re proud that we turned an ambitious idea into something playable: a full loop of placing collectors, linking them to factories, and routing track to keep up with growing demand.

What we learned

Thanks to MLH, we were able to work with Lovable.ai on the project after we had already started building it by hand. That gave us a chance to see where AI tooling speeds up development, like scaffolding interfaces and iterating on ideas quickly, and where human judgment still matters most, like game feel, balance, and design decisions.