Spot-Undoo: Privacy-First AI Kerala Micro-Travel Planner & Layover Explorer

Description and use

Spot-Undoo is a privacy‑first, local‑first micro‑travel planner for Kerala that uses locally hosted AI and MongoDB Atlas to generate constraint‑guaranteed itineraries on demand, catering to travelers and transit commuters.

Problem

Travel apps funnel users to overcrowded tourist traps, missing local culture and leaving transit commuters with no curated routes within time and budget constraints.

Outcome

Spot-Undoo delivers privacy‑first, locally hosted itineraries that expose authentic Kerala spots, support transit commuters with safe return guarantees, and enable community contributions via MongoDB Atlas.

Workflow

• 1. Open app – GPS auto‑snaps to nearest Kerala railway hub.
• 2. AI model (Qwen 2.5 3B fine‑tuned via Ollama) generates 3 plan archetypes based on user constraints.
• 3. Planner displays itineraries on Neo‑Brutalist vector canvas or Live OpenStreetMap.
• 4. User selects itinerary; app ensures safe return before train departure.
• 5. Community can upload new spots; MongoDB Atlas stores data with geospatial duplicate guard.

Evidence

Spot-Undoo solves this by pairing locally hosted open-weight AI models (fine-tuned Qwen 2.5:3B on Ollama) with MongoDB Atlas and a Native Android build to generate constraint-guaranteed itineraries on the fly.

Source

Runs locally using Ollama (spot-undo-qwen). ~71.4 tokens/second on an RTX 3050 mobile GPU (warm latency: 2.39 seconds).

Source

Built-in Geospatial Duplicate Guard (rejects identical spots within a ±0.00001 coordinate tolerance).

Source

Native Android APK & Responsive Mobile Web: Packaged with Capacitor 6 for native Android deployment (5.2 MB footprint).

Source