Open Source in Finance Forum Toronto 2026

FINOS AIGF provides a practical, use-case-led way to identify the risk profile of GenAI/agentic systems and select mitigations that are defensible for financial services. In parallel, the FINOS AI Reference Architecture Library is explicitly building threat-modelled architectures whose risks and mitigations are drawn from the AIGF catalogue, so teams can design, deploy, and operate within agreed risk tolerances.

This session proposes (and demos) an AIGF-aligned reference implementation: a composable, GitOps-delivered blueprint that turns FINOS AI reference architecture diagrams into deployable, secure platform patterns. Using k0rdent AI for consistent multi-cluster AI platform provisioning and Flux CD for policy-driven delivery and drift control, we operationalize the FINOS Multi-Agent Reference Architecture threat model (threats mapped to controls across layers such as gateway, MCP, evaluation, and observability). The result is a reusable stack that can be instantiated across cloud and bare metal, and that produces measurable assurance evidence aligned to Use Case → Risks → Metrics.