OPTIMESS

Inspiration

This idea came from something we experience every day. In hostel life, many of us want to get fitter or eat better, but when we go to the mess, we’re limited to a fixed set of options. Most of the time, we just guess what’s healthy and end up eating randomly. We realized the real problem isn’t lack of motivation—it’s that no one helps us choose the best possible option from what’s actually available.

What it does

We built a system that takes the daily mess menu and a user’s fitness goal, and then suggests what they should eat. We used AI to understand the menu and combined it with simple optimization logic to recommend the best possible combination. Our focus was to keep it intuitive and fast, so users can make decisions without overthinking.

How we built it

We built the application using Python and Streamlit to create a simple and interactive user interface. Users upload a mess menu image, which is processed using Anthropic Claude AI (vision + LLM) to extract food items for each meal.

Challenges we ran into

One of the biggest challenges was handling messy, unstructured menus and accurately interpreting different types of Indian food. Another challenge was ensuring that our recommendations were practical and realistic, not idealistic. We also had to balance functionality and simplicity while building everything within a limited timeframe.

What we learned

We learned that solving meaningful problems means working within real-world constraints, not ideal situations. Instead of assuming perfect conditions, we focused on what people actually face daily. We also saw how powerful AI can be when it simplifies small, everyday decisions rather than making things more complex.