Atlas Meridian is the fictional banking dataset for this model. It is synthetic, non-production, and not real bank data or customer telemetry.
Independent project · AI-assisted · verified build · synthetic data
My cyber-exposure architecture projectSecurity architecture · engineering · automation
I make closure
verifiable.
I developed Exposure Control with AI assistance to explore risk scoring, ownership, remediation, and validation. The data is synthetic. This is independent, non-production work.
Optional review tools
Explore the six-stage architecture
- SignalsFragmented findings
- Canonical ModelNormalized records
- Business ContextApps, assets, owners
- PolicyScored priorities
- OwnerOwner and team routing
- EvidenceValidated closure
SIMULATED EXAMPLE · EXP-1031
From public listing to validated closure.
Detect
analytics-archive-simPublic object listing
Route
Data PlatformPriya D.
Remediate
Removed the public policyadded a preventive control
Validate
Configuration diffanonymous-access retest
Without valid, current evidence, work stays open.
Explore the synthetic records100 applications. 1,000 assets. 20 teams.
I use Atlas Meridian, a fictional banking dataset, to model applications, assets, teams, ownership gaps, and source quality. It contains no employer or customer data.
- 100
- Applications
- 1,000
- Assets
- 20
- Engineering teams
- ~10,000
- Correlated exposures
Validated synthetic scale model; not a production benchmark. The dashboard makes all 10,000 synthetic records filterable and keeps 12 representative cases for readable detail.
From a finding to an owner and a validation result.
A source finding is normalized, enriched with business context, evaluated through policy, routed to an accountable owner, tracked through remediation, and only closes when current validation meets the closure conditions. A failed gate returns the work.
A real deployment requires separately permissioned identities, authoritative scanner retests, operating controls, and measured reliability.
Schemas, scoring, routing, and closure checks.
I used AI assistance to develop the lifecycle, schemas, scoring, adapters, ownership routing, persistence, validation gates, and tests. This project does not represent unaided coding expertise or a production client implementation.
Runnable reference implementation
Local-only CSV ingestion, deterministic risk scoring, accountable routing, a GitHub Issues adapter, two persistence adapters, repository scanners, a dashboard, and a synthetic lifecycle demo.
Authoritative sources and operations
IAM, EDR, CSPM, CMDB, Jira, and ServiceNow integrations are extension points for production integration. Production also needs recovery, security, reliability, and scale controls.
No silent assignment
Asset-to-application and application-to-team context resolve ownership. Missing, stale, or conflicting ownership is held for governance review and escalation instead of being routed silently.
Review the dashboard, model, and limits.
I use synthetic data so you can inspect the design without employer or customer information.
Working synthetic lifecycle
The public decision dashboard demonstrates the finding-to-closure model with synthetic systems, owners, remediation states, and validation records.
Fictional banking dataset
Atlas Meridian is a fictional, non-production enterprise context used to reason about applications, assets, teams, owners, source quality, and decision routing.
Quality gates completed
The portfolio build, type, lint, automated-test, dependency-audit, and browser checks completed successfully before publication.
Synthetic data and private source
Synthetic data demonstrates the design, implementation source stays private, and the dossier states what production adoption would require.
How I approach the work.
Use the project to discuss how I model findings, explain remediation priorities, assign owners, and check closure evidence.
