MUSE Evidence Support Platform
A generative AI platform for research evidence collection, causal-path construction, and impact tracking.
Technologies Used
Next.jsMastraGeminiRAGVercel
Key Features
Research evidence collection support
Causal-path construction
Impact tracking through impact certificates
Embedding and RAG quality improvements
Observability and evaluation for AI agents
Overview
MUSE supports evidence-based decision making by collecting research evidence, building causal paths, and tracking impact through impact certificates.
Role
Naoki led the platform build, including research-evidence embeddings, RAG tuning, and implementation and operation of observability and evaluation workflows for improving AI-agent quality.
Technical Focus
The work focused on more than generating answers. It connected evidence retrieval, causal reasoning, and impact tracking into a usable product workflow.
Completed on: Jun 1, 2025