Fraud Detection
Transaction risk, explained and queued.
A CPU-oriented API lab for asynchronous Nigerian transaction fraud detection and model operations.
A research system. Public runtime access is provided only when approved.
> system.status
ACTIVE_LAB
> capability.scan
ASYNCHRONOUS SCORING
REASONED RESULTS
MODEL OPERATIONS
NIGERIAN CONTEXTThe idea
The Fraud Detection lab explores asynchronous risk scoring for Nigerian transaction patterns. FastAPI accepts work while Dramatiq and Redis process jobs outside the request cycle.
The model work uses a five-million-row synthetic Nigerian transaction dataset with regional, channel, feature, and fraud-type variation. It is a technical demonstration—not financial, compliance, or legal advice.
Product system
Designed around the real loop.
Features are built to reinforce the core experience—not compete with it.
Asynchronous scoring
Submit a job, poll status, and retrieve risk output without holding the original request open.
Reasoned results
The API can return interpretable factors alongside a prediction workflow.
Model operations
Fine-tuning and model-status workflows support experimentation beyond a single fixed model.
Nigerian context
Synthetic training data includes region, channel, feature, and fraud-type variation designed around Nigerian transactions.
Answers / Fraud Detection
Good questions,
clear answers.
Can this approve or reject a real transaction?+
Not on its own. This is a lab and its output is not financial, legal, or compliance advice.
Is the training data real customer data?+
The documented training corpus is synthetic and designed to reproduce a broad set of Nigerian transaction characteristics.