AI-powered fraud detection platform with composable rule engine, LLM-generated investigation narratives, and real-time risk scoring across the full Detect-Investigate-Mitigate lifecycle.
RiskPulse is a full-stack SaaS platform that automates the fraud investigation lifecycle: Detect → Investigate → Mitigate. The system features a configurable rule engine supporting threshold, velocity, blacklist, pattern, and composite detection strategies. Cases are enriched with AI-generated narratives using OpenAI or Anthropic, with graceful fallback to local heuristics. The platform includes RBAC (Admin, Analyst, Engineer, Read-only), entity graph exploration, real-time dashboards with Recharts, global search, keyboard shortcuts, and a deployment pipeline with rollback and audit logging. Built on a monorepo architecture with Next.js 14 frontend, FastAPI backend, PostgreSQL, and Redis worker queues.
| Entry Point | apps/api/main.py / apps/web/src/app/layout.tsx |
| Build | docker compose up --build |
| Run | docker compose up |
| Architecture | Monorepo SaaS — Next.js frontend + FastAPI backend + Redis workers |
| Dependencies | 35 |
| Docker | Yes |
Monorepo SaaS — Next.js frontend + FastAPI backend + Redis workers