Congratulations! You have earned a place among the top 10% of candidates who made it through the selection process. This challenge is designed to test your creativity, problem-solving skills, and technical expertise.
However, with great talent comes great responsibility. This challenge will not be easy. It may seem complex at first glance, even overwhelming. It will demand time, focus, and innovation—even to fully grasp its scope.
But we know you don’t back down from a challenge. In fact, true talent emerges in the face of adversity. We are confident that you will prove yourself up to the task.
- Start Date: January 18, 2025
- End Date: January 19, 2025
Welcome to Cosmic Cycle 789, where humanity has expanded beyond the limits of the known universe and into uncharted dimensions. In this vast and intricate reality, gastronomy has evolved into an art form that transcends both space and time.
Interdimensional restaurants flourish across the cosmos:
- Sushi bars on Pandora serve Magikarp sashimi and Vaporeon dumplings.
- Tatooine taverns infuse dishes with Pipeweed for an enhanced flavor experience.
- High-tech eateries craft sauces using the enigmatic Slurm, blending contrasting flavors into interstellar delights.
However, with this expansion comes new responsibilities. The Galactic Federation strictly regulates ingredients, preparation methods, and certifications to ensure culinary safety across species. Chefs must navigate:
- Complex interdimensional regulations 📜
- Ingredients that exist in multiple quantum states simultaneously ⚛️
- Dietary restrictions for thousands of species across the multiverse 🌌
At the heart of this galactic cuisine is the Cosmic Pizza—a dish of legendary proportions. It is said that its mozzarella is crafted from the very essence of the Milky Way, and it requires the heat of three suns to bake. Some even worship it as a divine entity.
Develop an AI-powered assistant to guide intergalactic travelers through the vast and exotic culinary landscape of the cosmos.
Your system must be able to:
✅ Understand Natural Language Queries – Users will ask for dish recommendations in free-form text.
✅ Handle Complex Queries – Consider preferences, dietary restrictions, and cultural nuances.
✅ Process Multiple Data Sources – Extract and process information from menus, blogs, Galactic Federation laws, and cookbooks.
✅ Ensure Compliance – Verify dishes against galactic regulations.
Your system should utilize Generative AI techniques, including:
- Retrieval-Augmented Generation (RAG)
- AI Agents
- Multimodal Data Processing (text, images, structured data)
The AI must be capable of:
- Receiving natural language food requests
- Returning compliant and relevant dish recommendations based on the given documentation
🛠 Tech Stack:
The full list of dependencies is available in pyproject.toml
.
The solution consists of three core components:
-
Entity Extraction & Knowledge Graph Storage The ingeston pipeline parses documents to extract key entities such as ingredients, dishes, planets, and culinary techniques. The structured data is stored in a Neo4j graph database, allowing efficient query execution.
-
A LangGraph-powered agent that generates and executes Cypher queries Uses LLM-based reasoning to match user queries with structured graph data. The agent is empowered by a toolkit for accessing the Neo4j database, executing queries and returning parsed results.
-
Converts the query response into a structured output format Maps dishes to their respective IDs and handles cases where no exact match is found in order to align with the requirements of the Kaggle evaluation system.
The template is based on UV as package manager and Just as command runner. You need to have both installed in your system to use this template.
Once you have those, you can just run
just dev-sync
You can configure Ruff by editing the .ruff.toml
file. It is currently set to the default configuration.
Format your code:
just format
Run linters (ruff and mypy):
just lint