Our team developed an armed UAV entirely from scratch, and that was a challenging task. My job was even more challenging; I had to train a model that included at least 150k images, and the training process was very slow with Google Colab and Kaggle. So, I decided to train all the models locally, but first, I needed good hardware and perfect environment configurations for computer vision. For a while, I thought I would never finish the models and the dogfight algorithm before the competition deadline. But in the end, this adventure taught me one thing: even with limited timelines, if you don’t lose your passion, you will always find a way to improve your tasks. It might not be as perfect as you wish, but you can still make it.