A full-stack dining recommendation platform that uses user preferences, budget, social context, and intent to produce more relevant restaurant and experience suggestions.
Restaurant discovery is often generic. It usually filters by location or cuisine, but ignores why someone is going out, who they are with, what mood they want, and what constraints actually matter.
SAVR treats the dining decision like a context-aware recommendation problem. Users can describe their night, build their night through guided choices, or use a surprise flow to receive explainable recommendations.
Stack: FastAPI · React · TypeScript · PostgreSQL · JWT Auth · Recommendation Logic
Tradeoff
A simple restaurant list would be faster to build, but it would not capture why someone is going out, who they are with, budget, dietary limits, or atmosphere.
Outcome
Structured the product around Describe Your Night, Build Your Night, and Surprise Me so recommendation results can be explained and extended.
Tradeoff
Recommendation features become fragile when UI filters and backend scoring logic evolve separately.
Outcome
Kept API routes, data fields, and result cards aligned around explicit preference signals, restaurant metadata, and explainable output.
Tradeoff
Premium visuals can make a product look finished before the recommendation logic and data model are actually useful.
Outcome
Prioritized working flows, authentication, restaurant data, presets, APIs, and recommendation cards before final visual polish.
