This is a submission for the Built with Google Gemini: Writing Challenge
Few months back, NASA conducted its flagship hackathon named NASA Space Apps Challenge. As someone who has been passionate about space science since I was a little girl, there was no chance I could miss this. Gemini helped a lot in bringing my visualization to life.
What I Built with Google Gemini
There were a bunch of interesting problem statements. Out of these, I went with the problem statement titled "Meteor Madness". So, the challenge was to build a tool where you could simulate asteroid impact on our dear planet Earth along with recommending early preventive actions like Kinetic Impactor, Nuclear Explosion, etc. This one immediately caught my attention as earlier I had contributed to searching for unknown asteroids in a citizen science event (that's a story for another day).
First, I started with brainstorming what features this simulator should actually have. I did not want to build something that just throws numbers on the screen. I wanted users to feel the scale of the impact, understand the consequences, and most importantly see that there are possible solutions. I thought about having real asteroid selection, a custom scenario mode, live impact visualization on a 3D Earth, and a detailed impact report.
Once I had this rough idea in mind, I went to Gemini to refine and validate it. I discussed whether the features made sense together, whether the flow was logical, and how to structure the user journey from selecting an asteroid to seeing the final impact report. Gemini helped me organize scattered ideas into a more structured plan.
When I moved to implementation, especially while working with three.js, I again used Gemini to clarify concepts like scene setup, camera positioning, lighting, and raycasting for detecting the exact impact point on the globe. While integrating NASA NEO data processed through Python, I discussed how to efficiently pass meaningful parameters to the frontend and compute impact energy using proper formulas.
Whenever I faced debugging issues, instead of randomly experimenting, I would break the problem down and reason it out with Gemini. It helped me think more clearly and test solutions logically. I also used it to validate the scientific calculations for impact energy and seismic magnitude to ensure the simulator was not just visually impressive but also scientifically grounded.
Gemini was not building the project for me. It was more like a thinking partner that helped me refine ideas, validate decisions, and move forward with more confidence.
Demo
I have shared the complete working demonstration of The SkyWatcher here:
https://drive.google.com/file/d/1QIHVq4P335dWBqBbyyeoLFm0YTzPr5xf/view?usp=sharing
The live project can be accessed here:
https://nasa-space-apps-challenge-2025-umber.vercel.app/
In the demo, you can see both Real NEO Mode and Custom Sandbox Mode in action. You can select a real asteroid, analyze its threat level, choose a location on the 3D globe, and watch the full impact simulation along with the generated impact report.
What I Learned
This project taught me how powerful data becomes when it is visualized properly. Orbital parameters and velocity values might look intimidating in a spreadsheet, but when converted into motion, energy, and visible consequences, they tell a completely different story.
I learned how to integrate live API data using Python preprocessing and connect it smoothly to a frontend built with JavaScript and three.js. Handling real world scientific data made me more careful about accuracy. I understood the importance of grounding every simulation in real formulas instead of approximations just for visual effect.
Technically, I became much more comfortable with 3D rendering concepts, scene management, raycasting, animations, and performance considerations. I also realized that building something meaningful requires both technical depth and narrative clarity.
Most importantly, I learned that complex global problems can be made understandable with the right tools. Technology can reduce fear by replacing it with understanding.
Google Gemini Feedback
One thing I genuinely appreciated about Gemini was how straightforward it was. It did not sugarcoat mistakes or blindly agree with my ideas. If something did not make sense logically or structurally, the response reflected that clearly. That helped me improve the project instead of staying in a comfort zone.
What worked really well was brainstorming and refinement. When my ideas felt scattered, Gemini helped me organize them into a logical flow. It was especially helpful while validating scientific formulas, understanding three.js concepts, and debugging specific issues. It saved time by giving structured starting points instead of me staring at a blank screen.
Where I needed more support was in deeper 3D optimization and performance tuning. Some suggestions required multiple iterations before they worked smoothly in my setup. I also realized that the quality of output depended heavily on how clearly I framed my prompts. The more specific and detailed I was, the better the responses became.
Overall, Gemini worked best as a reasoning partner. It did not replace my thinking, but it strengthened it.
Top comments (0)