SwipeWork
Tinder-style job discovery app with AI summaries and fit scoring - swipe to save/skip jobs with smart matching.
Tech Stack
Key Highlights
Tinder-style swipe UI with card stack interaction
Supabase backend with RLS policies and auth
AI-powered job summaries in TR/EN languages
Hybrid fit scoring with embeddings + keywords
Offline-first with SQLite cache and sync
Privacy-focused with human-in-the-loop design
Project Details
A Tinder-style job discovery app with AI summaries and a fit score—swipe to save/skip, view structured details, and track leads. Currently migrating data/auth to Supabase and wiring the scoring pipeline.
**Status:** In progress (MVP) — core UI and data model drafted; Supabase (Postgres + Auth + Storage) setup in flight.
What's done so far:
**Data model & ingestion (draft):** Normalized job schema (title/company/location/salary/tags), near-duplicate detection plan (company+title+similarity).
**Swipe UI (Flutter):** Card stack with save / skip / super-like, detail drawer, quick filters scaffold (remote, salary band, stack); state mgmt with Riverpod/Bloc.
**Supabase groundwork:** Project created, Postgres schema WIP, Row-Level Security (RLS) policy draft, Supabase Auth (email/password) integration path identified; using supabase-flutter SDK.
**AI assist (prototype):** JD → 5-bullet summary (TR/EN) prompt templates; "top reasons to match" explanation format.
Up next (near term):
• Finish Supabase: Tables, indexes, RLS policies; implement saved/applied lists and on-device cache (SQLite) + sync.
• Fit scoring service: Standalone FastAPI microservice (Docker) for hybrid match (multilingual-e5-base embeddings + keywords) returning 0–100 fit + rationale; call from app.
• Notifications: New high-fit role alerts; rate-limit/backoff.
• Privacy & control: Human-in-the-loop only (no auto-apply), secure credential storage.
My contributions (so far):
• Designed the data model and drafted de-dup heuristics; built the Flutter swipe UI and state management.
• Set up Supabase project, started Postgres schema and RLS policy design; integrated supabase-flutter auth skeleton.
• Prototyped JD → AI summary pipeline (TR/EN) and match-reason templates.