RAG Knowledge Chatbot
Bilingual RAG-based knowledge chatbot for internal use across Japanese and English environments.
Role: AI Specialist Date: March 2025
PythonVLLMOllamaLlama.cppn8n
Problem
Internal knowledge was scattered across Notion, documents, and team members' heads. Finding relevant information required asking colleagues or searching multiple systems.
Approach
Built and deployed a RAG-based knowledge chatbot managing the full pipeline: embedding model selection, reranking strategy, text generation model evaluation, system prompt engineering, and guardrail configuration. Operated across both Japanese and English language environments.
Outcomes
Centralized internal knowledge access through a conversational interface. Significantly reduced time spent searching for information across departments.
2
Languages supported
Unified
Knowledge sources