Offline Adventures - lets start journey with AI

Description and use

Offline Adventures is a local‑first activity recommender that uses Streamlit and a local Llama model to quickly suggest activities based on time, energy, weather, and interests, ensuring privacy and deterministic constraints.

Problem

Users need a quick, privacy‑preserving way to find suitable local activities based on time, energy, weather, and interests.

Outcome

The system provides privacy‑preserving, deterministic‑constraint‑aware activity recommendations with semantic ranking, achieving 30 outdoor, 8 sheltered, and 12 indoor‑bridge activities in a curated catalog.

Workflow

• Collect user inputs (time, energy, weather, place types, group, interests).
• Apply deterministic Python filters for hard constraints (time, energy, weather, place types, group, candidate ID validation, session diversity).
• Select up to 15 approved candidates.
• Send candidates to local llama3.1:8b-instruct-q4_K_M via Ollama for semantic ranking and personalization.
• Validate structured JSON output with Pydantic.
• Present recommendation with mission and notes in Streamlit.

Evidence

Offline Adventures is a local-first activity recommender designed to make choosing an activity quick, then get out of the way.

Source

The application uses a deliberately hybrid architecture.

Source

Deterministic Python code handles hard constraints: available time; energy level; current weather; accessible place types; group type; candidate ID validation; session diversity.

Source

After filtering, up to 15 approved candidates are sent to a local llama3.1:8b-instruct-q4_K_M model through Ollama.

Source

The repository includes the curated 50-activity catalog, automated tests, a catalog validator, and a real Ollama smoke test.

Source

At submission time the project had 19 passing tests and a validated catalog distribution of 30 outdoor, 8 sheltered, and 12 indoor-bridge activities.

Source