Multi-Agent Carbon Fraud Detection System

Problem

Voluntary carbon markets cannot scale if every project claim and emission sheet requires manual review. Policy-aligned verification checks need to run in parallel across multiple data sources without overwhelming human analysts.

Approach

Built a multi-agent AI system with specialist roles, personality-based consensus agents, and adversarial validation. Integrated CLIMATIQ API, OpenWeather, and satellite data into a FastAPI services architecture deployed on Google Cloud with Vertex AI for ML workloads. Real-time fraud detection pipeline with PostgreSQL and Redis for state management.

Result

Verification time dropped 70% while catching more fraud. Analysts review flagged cases instead of reading every document line by line.

• ◆70% reduction in manual verification time
• ◆More fraud caught with fewer false positives
• ◆Analysts review flagged cases, not every document
• ◆Scales to handle parallel checks across emission claims