Canary Deployment with Automated Health Gating
Ships a .NET app to an Azure App Service canary slot, routes 10% of live traffic, then promotes or rolls back automatically based on Application Insights telemetry — no human in the go/no-go loop.
Five years of production Azure: I led an on-premises datacenter → Azure migration, built the CI/CD pipeline that took releases from ~2 hours to ~20 minutes, and implemented the cloud-side SOC 2 controls that pass an auditor's review. My focus now: compliance-grounded pipelines and private, policy-governed Azure landing-zone design for regulated environments.
The intersection most Azure DevOps candidates only cover half of: shipping pipelines and making them survive an audit.
Designed and executed an on-prem datacenter → Azure migration for production SaaS workloads — subscriptions, networking, Azure DevOps, and cutover. Now building private hub-and-spoke landing zones with private endpoints and policy guardrails.
Multi-stage Azure DevOps pipelines with canary slots, telemetry health gates that fail closed, and automatic promote-or-rollback. The go/no-go decision is made by data, not a human watching a dashboard.
SOC 2 cloud controls — Entra identity, conditional access, Defender, change-management gates — implemented so the pipeline produces the audit evidence instead of someone assembling it after the fact.
Clean-room builds and system designs — runnable repositories, not slideware. Each isolates one hard problem from production-grade Azure DevOps work.
Ships a .NET app to an Azure App Service canary slot, routes 10% of live traffic, then promotes or rolls back automatically based on Application Insights telemetry — no human in the go/no-go loop.
Treats a CVE-patched AKS node image as a promotable, gated artifact — staged, canaried in a tainted prod nodepool, regression-tested, PDB-protected on promotion, then verified against Defender — instead of an in-place --node-image-only upgrade that makes production the first place the new image runs.
A private-only hub-and-spoke design connecting Databricks, two AKS clusters, and three Container Apps environments — all PaaS via private endpoints, no public IPs, east-west forced through an Azure Firewall, blue/green cutover as a single Pulumi config flip.
A single stack that chains the mechanics that actually break real Terraform estates — backend state locking, for_each/dynamic, validated variables, import of out-of-band resources, refresh-only vs. apply drift handling, and a zero-diff module refactor with moved blocks.
Build artifacts are independent clean-room demonstrations of patterns I've shipped in production, isolated so the mechanics are easy to review. They are not the property of any prior employer. Repository links shared on request.
Workforce- and visitor-management SaaS for regulated and compliance-driven industries — offshore drilling, manufacturing, healthcare, and union-restricted facilities.
Re-establishing a current, legible baseline after the break — paired with the hands-on builds above so it's never "certified but never shipped it."
Open to senior Azure DevOps / platform engineering roles. If you're modernizing pipelines, standing up a private Azure landing zone, or prepping a SOC 2 audit, I can take ownership from day one.