AI Chatbot VEDEM
Internal conversational assistant built on a RAG architecture: it retrieves relevant passages from a document base and relies on a language model to write the answer.
Finding information in a large internal documentation takes time; an assistant should be able to answer based on those documents.
Internal AI assistant project carried out with the team's AI engineers.
A conversational assistant that first retrieves relevant passages (vector search over embeddings), then passes that context to a language model to write the answer.
Teamwork. The AI building blocks (RAG pipeline, model selection and tuning) are designed together with AI engineers. I handle application integration, the backend and production deployment.
The chat interface calls a backend API. The API retrieves relevant passages from a vector index, then queries a language model with that context.
User
Chat interface
Backend API
orchestration
Vector search
document embeddings
LLM API
answer generation