This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend.
What I Built
My brother is currently staying in a hostel, and one of the small problems he deals with is deciding what to cook.
When you're staying in a hostel, you usually don't have a fully equipped kitchen or a lot of ingredients. Sometimes you have a few things in your room, a small budget, and maybe 20–30 minutes to cook something.
So I thought, instead of searching for recipes every time, why not make something where you can simply enter what you already have and get some ideas?
That's how I came up with HostMeal.
HostMeal is a small AI-powered meal assistant that helps hostel students decide what they can cook based on their available ingredients and situation.
You can enter things like:
Ingredients you already have
Your budget
How much time you have
Number of servings
Dietary preference
Cooking equipment
Taste preference
Type of meal The app then gives you meal suggestions that match those requirements.
It also shows which ingredients are missing, provides the recipe, creates a shopping list, and finds a relevant YouTube cooking video.
The main idea is pretty simple:
Tell HostMeal what you have, and it helps you figure out what you can make.
I originally made it for my brother, but while working on it I realized that this could also be useful for other students living away from home.
Demo
Live Application: https://hostmeal.netlify.app/ Code
GitHub Repository: https://github.com/reyan3/HostMeal_Hacktoberfest01 The project has a React frontend and a Node/Express backend.
I've kept the project open source so that other developers can look at the code, suggest improvements, or contribute new features.
How I Built It
I used:
React for the frontend
Tailwind CSS for styling
Node.js + Express for the backend
Gemma for generating meal recommendations
Hugging Face for hosted model inference
YouTube Data API for finding recipe videos
Netlify for the frontend
Render for the backend
How it works
The user first enters their ingredients and preferences in the React app.
For example: Eggs, bread, onion
Budget: ₹80
Time: 20 minutes Equipment: Gas stove
The frontend sends this information to my Express backend.
The backend creates a prompt with these requirements and sends it to Gemma. I ask the model to return the meal information in a structured format so that I can use it directly in the frontend.
The response contains things like the meal name, ingredients, missing ingredients, and cooking instructions.
For the YouTube videos, I didn't ask the AI to generate video links. Instead, my backend uses the YouTube Data API to search for an actual recipe video based on the generated meal. So the basic flow is:
User
↓
React
↓
Express Backend
↓
Gemma
↓
Meal Suggestions
↓
React
↓
YouTube Data API
↓
Recipe Video
One of the more challenging parts was making the AI consider multiple things at once.
For example, a recipe might use the right ingredients but take 45 minutes to prepare when the user only has 20 minutes. Or it might require an oven when the user only has a gas stove. So I had to spend some time working on the prompt and the structure of the response.
This was also one of my first projects where I used an open-weight AI model as an actual part of the application rather than just experimenting with AI separately.
Why Does Open Innovation Matter?
For this project, using Gemma gave me the chance to work with an open-weight model and understand how it can be used inside a real application. I liked that I could experiment with the model and think about different ways of running it in the future instead of building the whole project around a single closed AI service.
There are also things I'd like to try later, such as using a smaller model locally, improving the recommendations for Indian hostel food, and adding more regional recipes.
Making the project open source also means I'm not the only person who has to decide where it goes next.
Someone could add a better recommendation system, improve the UI, add more meal options, or completely change how the recipe generation works.
That's one of the reasons I wanted to build this as an open-source project.
Prize Categories
Best Use of Render -
I'm using Render to deploy and host the Express backend for HostMeal.
Best Use of Gemma -
I'm using Gemma as the AI model that generates the personalized meal recommendations.
Why I Built It
I didn't start this project with the idea of building a huge platform.
It started with my brother being in a hostel and me thinking about a very simple problem:
"What can he cook with the stuff he already has?"
That was enough of an idea for me to start building.
I also wanted to use Hacktoberfest as an opportunity to build something that wasn't just another practice project, but something that could actually be useful to someone close to me.
So, this is HostMeal — a small project I built for my brother, and hopefully something that can help other hostel students too.