Recipe De-Blogger

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady

What I Built

Recipe De-Blogger is a lightweight web app that helps home cooks get to the useful part of an online recipe without wading through long personal stories, ads, and extra page content.

Paste a recipe URL and the app extracts the recipe into a cleaner, easier-to-follow format, including the title, ingredients, and cooking instructions. It can also scale ingredient quantities for a different serving size when the original serving count is available. The goal is simple: spend less time scrolling and more time cooking.

The project is especially useful for people who cook from online recipes and want a quick, practical view they can follow while preparing a meal.

Demo

Demo Video

Code

Repository Link

The project is built with a Python/FastAPI backend and a simple HTML, CSS, and JavaScript frontend. The backend extracts structured recipe data from recipe pages and uses a local language model as an AI-assisted fallback when needed.

How I Built It

I built Recipe De-Blogger with a small, practical stack:

• Python and FastAPI power the backend API.
• HTML, CSS, and JavaScript provide the frontend.
• Recipe-page extraction checks for structured recipe data, such as Schema.org recipe markup, before falling back to other extraction methods.
• Ollama runs an open-weight language model locally for AI-assisted recipe extraction, including extraction from available YouTube transcripts.
• Recipe scaling logic adjusts ingredient quantities for a target serving count when the original serving count is known.

I focused on keeping the app straightforward: provide a URL, extract the recipe, show the important details, and make serving-size adjustments easier. Because extraction can vary between websites and transcript quality, the result should still be checked before cooking.

Why Does Open Innovation Matter?

Open innovation made it possible to experiment with AI-assisted extraction without making every recipe request depend on a paid, closed AI API. Running a model locally through Ollama gives me more control over the development process and lets me learn how an open-weight model behaves on real, messy recipe content.

It also keeps the architecture approachable. The app combines standard web technologies with a local model, so the extraction workflow can be inspected, tested, and improved as the project grows. For a small open-source project, that flexibility is valuable: I can iterate on prompts, parsing, and validation without tying the whole product to one proprietary model provider.

The Lazy Payoff

The payoff is fewer unnecessary steps between finding a recipe and starting to cook. Instead of repeatedly scrolling past a blog's introduction and other page content, users can get the recipe details in one organized view.

I haven't included a measured time-saving claim because I haven't run a formal timing test. The intended benefit is reducing friction: copy a recipe link, extract the key details, and focus on cooking.

Other Team Members

Rishi Raj Goel- @rishi_rajgoel_c5e99c8b81