It was late in the CS lab, and I was staring at my terminal trying to figure out why my custom shell wouldn't update when I hit backspace during an I/O redirection test. But the real frustration wasn't coming from my buggy code, it was coming from Matthew.
Matthew sat at the desk next to me, completely ignoring his lab work. Instead, his screen was split between the campus dining portal and a massive, chaotic spreadsheet. He was muttering to himself while aggressively punching numbers into his phone calculator. He had a strict daily goal of 3,000 calories and 60 grams of protein, and he was trying to manually piece together three realistic meals across Wiley, Windsor, and Ford.
He would find a great lunch at Earhart, write it all down, and then realize it left him with an impossible mathematical gap for dinner. That meant he had to scrap the entire plan and recalculate his whole day just to avoid eating nothing but plain chicken breasts and six servings of green beans at Hillenbrand. It was a classic constraint satisfaction problem, and he was stubbornly trying to brute-force solve it by hand every single night.
Watching him waste an hour doing manual arithmetic instead of writing his C code was the lightbulb moment. I told Matthew to close his spreadsheet. If we could build a shell from scratch, we could definitely write a quick script to automate his diet. By pulling the dining court dataset and running a randomized search algorithm against his daily targets, we could instantly generate a perfect meal plan for him in milliseconds. That is how the idea for the project was born.
Log in or sign up for Devpost to join the conversation.