01Life Time Coaching EcosystemEnterprise
Three clients, one truth. A member app, a trainer app on iOS and Android, and a web console all writing to the same programs — and a trainer changing a workout on the floor has to land on the member’s phone before they finish the set.
Built and maintain the coaching platform across all three surfaces. React and Node on top, MongoDB and Azure underneath, Kubernetes holding it up.
02Monolith → MicroservicesInfrastructure
Cutting a live system apart without the members noticing. Every seam is a place data can go missing, and there was no maintenance window worth taking.
Led the migration to 50+ independently deployable services with their own CI/CD and autoscaling. Deploys went from 15+ minutes to under 5, per service, in parallel.
03L.AI.C — Coaching AssistantAI / ML
Generation was never the hard part. A suggestion that ignores the methodology a coach already trusts gets overridden — and an assistant that gets overridden is one nobody opens twice. Fitting the way trainers actually work on the floor was the job.
Built the trainer-facing assistant inside L.AI.C, our AI coaching platform — the surface a coach actually works in when they build a member’s program.
04Kafka Event PipelinesInfrastructure
Ordering and idempotency. Fifty services, at-least-once delivery, and a member who must never see the same session logged twice.
Designed the producers, consumers, and topic structure carrying real-time data across the service mesh.
Nobody assigned this one. Co-founded the LLC, chose the stack, and shipped it around a full-time job — which means every architectural shortcut is a bill I pay personally.
Family and business management app. I run the backend: a .NET C# API and a Python/Node MCP server on FastMCP, behind a Kotlin Multiplatform client.