Inspiration
Every Monday, our team gathers at the local Shamrock Inn. As broke college students, their 50-cent wing night isn’t just a cheap dinner, it’s a cherished weekly tradition that brings us together without emptying our wallets. From those wing nights to splitting BJ's weekly Pizookie specials, we realized something important: sharing food builds community, but budget constraints often stand in the way. We wanted to create Deals4Meals for people just like us, students, friends, and families who deserve to enjoy meals that offer both incredible quality and unbeatable quantity.
What it does
Deals4Meals is a crowd-sourced, community-driven platform where food lovers unite to share the absolute best local promotions. Whether it's a flash promotion like "Free Chicken Sandwich Day" or a legendary long-standing special like our favorite Shamrock Inn wings, users can easily post and discover deals in their area. But we didn't stop there. To truly help our users stretch their dollars, the platform features an AI-powered meal planner that intelligently builds a customized weekly menu using these active community deals, taking the stress out of budgeting for the week!
How we built it
The heart of our engineering approach was heavily inspired by SWEN 261, a class that taught us the profound value of building full-stack applications to serve the community and non-profit organizations.
To turn our vision into reality quickly, we fed our initial wireframes into Grok and Cursor. These tools helped generate the foundational boilerplate templates, allowing us to dive straight into the complex logic and give the application its true functionality. We built a lightning fast frontend using React and Vite, powered the backend with a robust custom API, and managed our data using Tiger Data. To bring our automated weekly planner to life and user input verification, we integrated the Google Gemini API. Finally, we tied it all together by manually deploying the entire ecosystem onto a Vultr VPS.
Challenges we ran into
Bringing this platform to life was a true test of our perseverance. Setting up the infrastructure from scratch on a Vultr VPS proved surprisingly difficult, complete with unexpected billing loops and steep learning curves. As our vision for the platform's potential kept expanding, we outgrew our own database schemas multiple times, requiring us to completely rework our data architecture on the fly. Additionally, with four developers working simultaneously in a monorepo, we inevitably ran into overlap and minor misunderstandings that required us to frequently step back, communicate, and untangle our code.
Accomplishments that we're proud of
We are incredibly proud to have transformed a hackathon idea into a fully functional, 24/7 live product. Seeing our application successfully deployed on a VPS with complete, stable database support feels like a massive leap forward for all of us. While the product is very much a prototype, it is a good base for what could be and we are excited to keep growing it!
What we learned
We learned exactly what it takes both technically and financially to host and run real world infrastructure. Connecting multiple distinct providers and ensuring they communicated flawlessly was a massive undertaking. When our custom APIs initially failed to reach the database, we learned how to trace network requests, monitor ports, and identify the exact moments our operations dropped.
What's next for Deals4Meals
Our next steps include:
- Growing a dedicated community: Bringing in more users to share the love (and the deals) across new cities.
- Long-term maintenance: Keeping the VPS architecture stable, secure, and scalable.
- Mobile Port: Turn Deals4Meals into a mobile app for ease of access
- Enhanced user profiles: Building a better account system so users can save their favorite deals and customize their dietary preferences.
- Lightning-fast AI: Optimizing our Gemini prompts and backend operations to deliver personalized meal plans even faster.
Built With
- css
- cursor
- express.js
- gemini
- github
- grok
- pnpm
- react
- tigerdata
- typescript
- usdafooddata
- vite
- vultr