Building AFLD — Autonomous Financial Leak Detector (AI-Driven Revenue Recovery)

Hey community! I’m Rocío, Data Engineer & Cloud Architect. I’m currently building AFLD (Autonomous Financial Leak Detector), an agentic workflow designed to stop invisible revenue leakage in real time before it settles.
The Problem:
Traditional audits catch leaks too late. Static rules miss edge cases across massive data streams, costing companies millions in lost revenue.
The Tech & Innovation:
AFLD leverages Claude 3.5 Sonnet on Amazon Bedrock paired with the Model Context Protocol (MCP) for context-aware auditing across S3 data lakes and Redshift warehouses. Powered by Apache Flink for stateful streaming, it doesn’t just flag anomalies—it reasons through live business context in milliseconds.

Where we are & What we need:
Status: Production-ready core architecture deployed via Terraform.
Goal: Raising pre-seed / connecting with cybersecurity & data infrastructure VCs, advisors, and design partners (Fintechs/Enterprises processing streaming transactions) to pilot and scale the product.

Would love to connect with anyone building in AI Governance, Data Infrastructure, or Fintech Risk!