Rebuilding customer support at Luxury Escapes

The problem

Support ran across a legacy stack and a new one, and tickets were routed by hand. The team needed to migrate without disrupting live support, route tickets faster, and ship changes more safely.

How it fits together

1. 01Zendesk & GenesysNew tickets and calls raise events.
2. 02SQSEvents queue up, schema-validated and rate-limited.
3. 03Lambda + OpenAIA classifier labels intent and urgency.
4. 04Routing rulesLabels map to the right queue and team.
5. 05Support agentsThe ticket lands with someone who can help.

Alongside it, a two-way sync kept the legacy and new stacks consistent until the switch-over.

What I did

• Designed OpenAI-based classification services integrated with Zendesk and Genesys for real-time ticket routing.
• Built event-driven microservices on SQS and Lambda with schema validation and Valkey-based rate limiting.
• Implemented a two-way sync between the legacy and modern support stacks so both could run during the migration.
• Led the front end of the new Support Platform, delivering 12+ releases under tight deadlines.
• Built a shared internal library used across services to standardise integration patterns.
• Set up Jest, Cypress and CircleCI pipelines, with LocalStack and TestContainers for realistic integration tests.
• Acted as primary reviewer for all code across the platform's repositories.