Detect
Transaction, AML, sanctions, and KYC signals are normalized into actionable business alerts.
Compliance engineering
Einhaltung is an event-driven platform that brings transaction monitoring, investigations, sanctions, KYC, approvals, and analyst work into one governed compliance domain.
The challenge
Compliance teams must investigate high-volume transactions, KYC reviews, sanctions matches, and approvals while preserving every decision and its evidence. Tightly coupled workflows make those changes risky and make an audit trail difficult to reconstruct.
The goal was a dedicated compliance domain that can evolve independently from core financial-service systems, without compromising traceability or operational speed.
The solution
Transaction, AML, sanctions, and KYC signals are normalized into actionable business alerts.
Cases hold assignment, notes, escalation, evidence, and approval state in a single analyst workflow.
Immutable audit events and reporting projections make each action visible, searchable, and reviewable.
Architecture
LeCore integrates through approved APIs and versioned events, keeping compliance records and decisions out of core schemas.
Write-side services protect transactional state. Reporting consumes domain events into read projections, so dashboards and search do not overload operational systems.
Choreography-oriented sagas make long-running states visible. Idempotent consumers, business keys, unique constraints, and transactional outboxes make retries safe.
Trust by design
SecurityThe gateway validates LeCore sessions and roles; workloads use restrictive browser headers, default-deny networking, managed secrets, and non-root execution.
ObservabilityHealth, Prometheus metrics, OpenTelemetry traces, and Grafana dashboards surface availability, throughput, and resource health.
DeploymentHardened Kubernetes manifests add health probes, disruption budgets, autoscaling, topology spread, and production preflight checks.
Foundational integration roadmap
The first release is advisory: Einhaltung detects and investigates risk after a LeCore event. It does not hold or release money until that control path has been separately designed, tested, and approved.
Publish a backward-compatible LeCore payment event with sender, beneficiary, account, channel, jurisdiction, correlation, and idempotency context. Einhaltung consumes both the existing contract and the enriched version during rollout.
Consume the event idempotently, evaluate one high-value-transfer AML rule, normalize the result into an alert, and create an assigned investigation case with the source evidence attached.
Publish a versioned Einhaltung business-alert event. VectaSinq delivers an in-app and email notification to the compliance queue, with retries, delivery history, and no direct database coupling.
Phase 1: LeCore transfer to Kafka to Einhaltung transaction monitoring.
Phase 2: AML rule to alert, case, audit event, and analyst-workbench queue.
Phase 3: Einhaltung business-alert event to VectaSinq, then analyst email and in-app delivery verification.
Phase 4: Contract, replay, duplicate-delivery, failure, and end-to-end tests across all three repositories.
LeCore owns money movement. Einhaltung owns risk decisions, evidence, and compliance workflow. VectaSinq owns templates, channel selection, retries, and delivery records.
All cross-platform communication uses authenticated APIs or versioned Kafka events. No service reads another platform's database.
After the foundation
Once the foundational LeCore and VectaSinq flow is proven, the platform can mature from a high-value alert workflow into a broader, governed compliance capability.
Versioned watchlists, fuzzy-match review, document evidence, periodic KYC refresh, and maker-checker approvals.
Velocity, structuring, geography, counterparty, and network rules with rule simulation, tuning, and evaluation history.
Elasticsearch investigation search, SLA and alert trends, regulatory reporting packs, and governed exports.
Separately approved hold and release APIs, explicit fail-safe behaviour, dual approval, and a complete audit trail.
Kafka replay and dead-letter operations, load testing, restore drills, threat modelling, penetration testing, and incident runbooks.
Analyst-approved evidence summaries and next-step suggestions, introduced only with explainability and governance controls.
Outcome
Einhaltung makes the trade-off of eventual consistency explicit: analysts can see workflow state, retries, approvals, and evidence rather than relying on opaque distributed transactions.