The GENIUS Olympiad was held at the Rochester Institute of Technology (RIT), USA. Upon arrival, we received our identity cards and were allocated our rooms for the duration of the event.

The following day, we were given a designated time slot to set up our projects. I began preparing my exhibit by inflating the pool we had brought all the way from India, carefully arranging the space to accommodate it.

In the evening, I rehearsed my presentation 3–4 times and ran the Python script integral to my project. However, I encountered unexpected technical issues. I spent the night troubleshooting—working tirelessly, attempting to fine-tune the code.

The primary challenge was that the machine learning algorithm, which performed well on a larger device in natural water bodies, was not functioning as expected in the smaller inflatable pool. The system was mistakenly identifying the pool’s boundary as plastic waste. I sought assistance from Claude AI several times, trying to refine the model in real time. Despite these challenges, I adapted my approach and decided to demonstrate the functionality of the code while explaining the reasons behind its limitations in this modified setup.

Judging and Interaction

Judging began promptly at 9:15 AM once all projects were in place. Over the course of three hours, three judges visited my booth. The experience was exciting and enriching. The judges were attentive and patient, allowing each participant the time to fully explain their work.

Key Learning

This experience taught me that innovation is not just about having a working model—it’s about adaptability. Real-world deployment environments can vary drastically from controlled settings, and a truly robust solution must account for such variability. It also reinforced the importance of resilience. Even when things didn’t go as planned, communicating the challenges honestly and focusing on problem-solving made a difference. Sometimes, improvisation and clarity in explanation are just as important as technical perfection.