Automated AML Transaction Monitoring

I’m building a transaction-monitoring platform that pinpoints suspicious activity across bank transfers and other payment rails in real time. The core requirement is an engine that ingests structured payment data, applies configurable rules and machine-learning models, and raises well-scored alerts the moment a transaction or account pattern looks off.

Key functions I need to see working end-to-end:
• Real-time detection of anomalous or high-risk bank transfers and payments, with adjustable thresholds and risk parameters
• Integrated sanctions / PEP screening and negative-news checks that run in parallel with the monitoring logic (existing API keys will be supplied)
• Alert dashboard with search, case notes, and escalation workflow so compliance officers can investigate, add evidence, and close cases
• One-click generation of compliance reports (SAR/STR-ready PDF or XML) drawing directly from case data
• REST or gRPC endpoints so the engine can be fed transaction batches and push back alert objects to downstream systems
• Role-based access, full audit trail, and encryption of data at rest and in transit to satisfy AML regulatory expectations

Stack and hosting are flexible; I’m currently leaning toward Python (FastAPI), PostgreSQL, and either on-prem Docker or AWS ECS, but if you have a proven alternative that delivers low-latency processing I’m open to hearing it.

Acceptance criteria
1. A working prototype processing a sample dataset of at least one million transactions at ≤200 ms per record.
2. 95 % test coverage of rules and alerting logic.
3. Documentation covering system architecture, rule configuration, and deployment steps.

When you reply, focus on the experience you have with AML, fraud detection, or high-throughput data streams—links to previous platforms or brief case studies are perfect. Feel free to outline a phased roadmap if that helps frame the build.