# DineTuned

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Rohit Sharma](https://cerebralvalley.ai/u/roshar29)
- **GitHub:** https://github.com/roshar29000-ds/DineTuned-OpenSource
- **Demo video:** https://drive.google.com/file/d/1buOEbuX-TYIbjsvw9RD9CRqwz9q8ukzn/view?usp=sharing
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3637

DineTuned is an AI-powered personal dining concierge designed around a problem that restaurant discovery platforms rarely solve: 
finding a restaurant is one decision; knowing what to order is another.

Restaurant recommendations are typically driven by ratings, popularity, cuisine, or location. But once someone opens a menu, those signals offer little guidance about which dish they personally will enjoy. Two people at the same restaurant can have entirely different ideal orders.

DineTuned approaches this as a problem of understanding individual taste and applying that understanding to real dining choices.

The experience begins with Taste Detective, an interactive preference-discovery system that investigates not only what users enjoy, but why: flavors, textures, preparation methods, spice tolerance, familiar dishes, and aversions. These signals form a seven-dimensional Taste DNA, an evolving preference profile that distinguishes established preferences from inferred characteristics and areas of uncertainty.

When a user selects a restaurant, DineTuned structures its menu into dish-level information, including descriptions, categories, prices, and available dietary attributes. Meta-powered culinary reasoning evaluates these dishes against the user's Taste DNA to identify strong personal matches and generate explanations grounded in their preferences.

Our signature feature, TasteBridge, connects unfamiliar dishes to established preferences across cuisines. Someone who enjoys smoky, aromatic preparations may discover a suitable dish from an entirely different culinary tradition because of shared sensory characteristics, rather than cuisine labels alone.

The experience extends beyond recommendations. Complete My Table suggests complementary dishes, budget-aware alternatives provide flexibility, and interactive refinements allow users to adjust recommendations. Through Taste Memory, users can save selections, record what they tried, and provide feedback.

Built with Next.js, FastAPI, Meta Model API, Google Places, and Supabase, DineTuned combines personalized culinary reasoning with deterministic controls for dietary restrictions, menu evidence, and pricing.

Our goal is to make ordering unfamiliar food feel less like guessing and more like receiving a recommendation from someone who genuinely understands your taste.

Live website: https://www.dinetuned.com

## More from Meta Global AI Developer Hackathon

- [MuseGuard](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3634)
- [Saath (Sticking Together)](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3635)
- [The Virtual Neuro Lab](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3636)
- [Ereditas](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3638)
- [Ekos Cyber Labs](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3639)
- [ForgeFront Systems](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3640)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3637. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt