Chwimiro (취미로)

Period: 2026
SeSAC mini project
Code: github.com/jeseo-lab/sesac_mini_project

Overview

Chwimiro uses a short survey and optional free-text input to understand a user’s constraints (time, budget, energy, and sociability). It then “prescribes” hobbies, together with a concrete first action and a path to get started.

My role

  • Planning: Designed a low-pressure onboarding, a 9-axis survey, prescription-style recommendation cards, “unexpected” recommendations, one round of feedback-based re-recommendation, hand-off to external platforms, and key metrics.
  • Back-end and AI: Built FastAPI recommendation and feedback flows; designed a hybrid tag + embedding + LLM recommender; designed ChromaDB and Supabase data structures; planned RAG, sensitive-information handling, and safety branching.
  • Front-end: Helped implement the step-by-step survey, recommendation cards, feedback and re-recommendation, and action-path screens, including responsive, loading, and error states.

Tech stack: FastAPI, LLM, embeddings, ChromaDB, Supabase