A conversational cooking assistant built using AWS PartyRock (Amazon Bedrock), developed as part of the Future AWS AI Scientist Track (AWS & Udacity Scholars Program).
🔗 Live Demo: PartyRock App
Traditional recipe apps return static, generic results. Cuisine Compass uses prompt-engineered LLM chains to generate personalized recipe suggestions, nutritional insights, and ingredient substitutions based on user-provided ingredients and dietary preferences — with no backend infrastructure required.
- Recipe Suggestion Engine — Generates tailored recipes from user-provided ingredients or a specific meal request
- Dietary Personalization — Supports multiple preference profiles: standard, keto, and diet-friendly
- Kitchen Help — Interactive Q&A for cooking tips and ingredient substitutions
- Dynamic Input Handling — Free-text ingredient entry + structured meal-type selection
Built entirely on AWS PartyRock's no-code generative AI framework:
- Input Widgets — Capture ingredients (free text) and meal type (dropdown)
- Prompt Chain 1 — Recipe Suggestion — Structured prompt generates a recipe with dietary properties and cooking tips based on user input
- Prompt Chain 2 — Kitchen Help — Secondary prompt chain handles follow-up cooking questions and ingredient substitution requests
| Component | Technology |
|---|---|
| AI/LLM Platform | AWS PartyRock (Amazon Bedrock) |
| Core Technique | Prompt Engineering / Prompt Chaining |
| Deployment | AWS-hosted, no-code |
This project focused on rapid AI prototyping — designing effective prompt structures instead of custom ML pipelines, and understanding how generative AI tools can deliver personalized recommendations without traditional model training.
Developed as part of the Future AWS AI Scientist Track, AWS & Udacity Scholars Program.