Inspiration
The main inspiration for this project was the frustration with not getting the correct brew every morning. I usually wrote everything down in my phone's notes app; however, it took time and was tedious. Also, knowing what changes to make when your brew is sour can be quite difficult at first, especially when brewing special coffee, which is very lightly roasted. So I always wanted an app that could ask AI how to adjust coffee and keep all my recipes in one place. It’s basically my coffee sidekick, no more guessing, no more sour surprises, just a way to finally nail that perfect cup and learn from people who’ve already figured it out.
What it does
Roast and Found is your coffee sidekick. It is the app I keep in my pocket, so I always have a good starting point before I’ve had my first cup. I use it to log every brew I make, from that perfect pour-over on a Saturday morning to the espresso shot I somehow nailed on a Monday morning, so instead of going through my Notes app scattered with my coffee notes, everything lives in one place. But it doesn’t just store recipes; it actually helps me figure out what I’m working with. I can snap a photo of my beans and get a roast level estimate, which is huge because I buy a lot of light-roasted specialty coffee that looks almost identical to medium roast, and knowing where it sits on that spectrum completely changes how I grind and brew it. Another major help is being able to ask AI about how to adjust my roast. If my coffee is too sour, I can ask it to suggest changes, or if there are no tasting notes in the coffee, it will also then suggest changes. The social feed is also a huge help; it is where I can see exactly how other people are brewing their coffee and what ratios they’re using. When someone posts a recipe that looks incredible, I just hit copy, and it saves straight to my own collection, so I can skip the guesswork and brew their winning setup myself.
How I built it
I built this project using SwiftUI for the frontend, running natively on iOS with Swift. The backend is a Python REST API that exposes endpoints for the AI features. It queries Google’s Gemma model for brewing adjustments and recommendations and uses the Google Cloud Vision API to estimate roast level from images by analyzing bean colour and texture. Data persistence and user management are handled through Firebase Firestore and Firebase Authentication.
Challenges I ran into
The smaller hurdle was getting Firebase configured correctly, just the usual dance of plist files and API keys. The real headache, and honestly the biggest challenge of the whole project, was a UI glitch in the social tab where the post image was overlapping the copy button. I spent three solid hours debugging it, tearing my hair out trying to figure out why the button worked fine in previews but refused to respond in the actual app. Of course, I finally cracked it right at the bitter end of the hackathon around 3:50 AM.
Accomplishments that I am proud of
I feel I am most proud of how much I was able to get done as a solo hacker, given that the app is small and only has a handful of features; however, I am still proud of being able to produce a usable and functional product.
What I learned
My biggest takeaway from this project is how much fun you can have developing something. I came into this hackathon not really even sure if I should go, but I have just had so much fun doing full-stack development and making a product I would use every day.
What’s next for RoastAndFound
The main thing that is next for Roast and Found is implementing more features, polishing up the app a little more, and getting an MVP that I can advertise on social media and have people enjoying their coffee a little more every day.
Log in or sign up for Devpost to join the conversation.