Woorinara Service
Woorinara is an application that provides information about administrative services for foreigners living in Korea.
Why did we make Woorinara?
Pain points
- Many foreigners struggle to understand the process and requirements of visa services. However, government websites do not update their English content frequently, and most details are only available in Korean.
- When visiting an immigration office, they often have difficulty understanding legal terms on application forms.
- Immigration officers usually speak only Korean, making it difficult to ask questions.
Interviews with Foreign Residents
- “I was confused when I visited the immigration office because I didn’t know which documents to bring.” (Brendon, USA)
- “The staff at the immigration office speak only Korean, and they speak very fast.” (Anahi, Argentina)
- “I was rejected many times because I filled out the forms incorrectly before submitting them.” (Aylin, Germany)
What Woorinara help you
- AI Chatbot - Free to ask about administrative services for foreigners.
- Forum - Share tips and experiences related to civil services.
- Translation and Form Examples - Provide translations of immigration office documents and auto-completes forms based on user information.
Roles and Responsibilities
My role : AI Engineer
- Developed the AI chatbot
- Crawled and processed data from government websites and loaded it into a database
- Designed the RAG (Retrieval-Augmented Generation) chatbot architecture
- Improved chatbot performance using user data
- Technologies Used: : Python, FastAPI, PostgreSQL, Qdrant, LangChain, OpenAI API, Docker, AWS Lambda
Team member (7 people)
- Product Manager, Marketer, Designer
- Backend Developer, Frontend Developer (iOS), 2 AI Engineers
About the AI Chatbot
- Provides information about administrative services for foreigners, including:
- Visa applications, extensions, changes, and residence modifications
- Offers location-based guidance for government agencies, including:
- Immigration offices
- Self-Service Civil Document Kiosks
- Foreign Civil Affairs Agencies
Architecture
- Client: Asks a question
- Backend: Verifies user ID & sends it to AI
- AI: Generates an answer
- Backend: Sends the answer back
- Client: Receives & checks it
RAG (Retrieval-Augmented Generation) Chatbot
- Retrieves relevant information from the database and provides it to the LLM (Large Language Model) as context for the user's question.
- External Databases:
- Location guide → Extracts location data from a relational database (PostgreSQL).
- Common Questions → Retrieves civil service data from a vector database (Qdrant).
- Challenges:
- Reducing response time
- Improving chatbot accuracy
- Integrating the chatbot with KakaoTalk
Acomplishments
🏆 [2nd Prize] Won funding in the Data Usage Competition
Issued by Korean National Information Society Agency (NIA) · Dec 2024
