Skip to content
ROTHERT.CC
SRCprojects/regulated-ai-platform.mdback to projects

Regulated AI Platform

A regulated gambling operator's core internal AI platform, built on AWS under GDPR and Dutch gambling regulation. Bedrock AgentCore provides the agent runtime, memory and identity layer. On top of it: RBAC-controlled RAG assistants and knowledge-base tools with human review, web crawlers and sanitisation checks; a copy-compliance system driven by custom-trained sequence-tagging models (GECToR-style) that makes up to 200 edits per campaign, checking text against internal policy and gambling law; ERP-integrated batch-payment checks with digital signatures and audit trails; player-welfare risk assessment over highly sensitive personal data; and automated legal screening for new betting markets. The VPC networking and ingress/egress controls that keep regulated data inside the compliance boundary are defined in Terraform. Worked to AWS best practice, with the architecture reviewed by the client's AWS-partner engineering team. The client is not named here.

see also

described here because the source is client-owned or closed
specvalue
languagePython
sourceclosed / client-owned

topics

  • AWS Bedrock
  • AgentCore
  • RAG
  • Terraform
  • GDPR