FindFest : oTP team 1

FindFest: Integrated Cultural Event Exploration Platform for Seoul

FindFest is a web service that consolidates scattered festival and event information from across Seoul into a single platform, allowing users to explore the city as if they were on a guided journey.

1. Project Background & Objectives

• Problem Definition: Addressed the difficulty of searching for events due to fragmented information across various channels (district offices, SNS, etc.), the lack of personalized management systems, and poor user experiences.

• Solution: Built an integrated platform providing city-wide data by incorporating an AI chatbot, user authentication, and a "Save/Favorite" feature.

• Target Audience: Primarily targeting young adults in their 20s and 30s with high demand for cultural activities, social gatherings, and extracurricular experiences.

2. Core Technologies & Implementation

Service Architecture

• Frontend: Utilized Next.js (App Router) and TypeScript to ensure fast loading speeds and data type stability.

• Backend: Developed with FastAPI and SQLAlchemy; managed relational data and vector embeddings using PostgreSQL and pgvector.

• AI: Implemented advanced search and chatbot functionalities using Upstage’s Solar LLM and embedding models.

• Deployment: Deployed on Amazon EC2 using Docker Compose for container orchestration.

Key Features

• Data Pipeline: Periodically collects the latest information via the Seoul Open Data Plaza API, performing de-duplication and vector embedding before storing it in the database.

• AI Chatbot: Utilizes a LangGraph-based workflow to analyze user intent (general conversation, recommendations, date-based searches, etc.) to provide intelligent responses.

• Personalization System: Implemented JWT-based authentication for secure login, allowing users to "save" events and manage them on a dedicated profile page.

• Dynamic Filtering: Provides a UI that allows users to instantly explore events through categories, date pickers (calendar), and keyword searches.

3. Collaboration & Achievements

• Collaboration Tools: Followed a GitHub branch strategy with PR-based (Pull Request) code reviews; managed schedules and meeting minutes via Notion.

• Key Results: Successfully integrated over 4,200 event data points and built a high-fidelity MVP (Minimum Viable Product) featuring both AI-driven interaction and personalized user management.