Custom contact-center applications, a Git-backed CI/CD pipeline for the platform, deep multi-system integrations, and an AI ops stack I built to run Genesys itself. At enterprise scale, handling roughly 900,000 conversations a month. The work is below.
40,000+ lines of production software: applications embedded in Genesys where the native client fell short (a real-time call-control desktop, a back-office inbox, agent-assist tools, a no-code analytics builder, a training app), plus the automation and AI tooling that runs the platform.
An LLM ops stack that runs the platform under deterministic guardrails, a speech-reactive IVR regression harness, vision-powered flow scoping, and a Git-backed flow-as-code CI/CD pipeline that turns a click-ops org into a version-controlled one.
Conversational-AI agents (Sierra AI, Kore.ai), CRM and ticketing (Salesforce, ServiceNow, Zendesk, Jira), EHR and HRIS (Epic, Workday), and cloud services (Azure Functions, AWS), wired into live contact-center flows. Platforms owned end to end, from dev to QA to production, and architected to scale.
Real production builds at enterprise scale. Filter by discipline, or by standard platform mastery vs. custom, out-of-the-box systems.
Not a claim, a log. Twelve months of platform authorship pulled straight from the audit trail: what was built, where, and how much.
Verified backup before any change, read-only credentials by default, hard gates on destructive and production writes, and an append-only audit log over every automated action.
Pre-flight validation that halts a batch rather than corrupting a campaign. Bounded, non-hanging retry on every input. A partial run can never close its own ticket.
State-tracked scheduling that cannot double-fire. A job can run, retry, or overlap a prior run without ever repeating or corrupting work, so reruns are always safe.
Behavior driven by data tables, so a routing, location, or alerting change is a row edit, not a republish. Adding a location or a department is one row.
Logic lives in Data Functions (custom JS/Node) and shared typed modules, not stitched together as low-code data actions. Architect is just the trigger surface, and other engineers' flows consume my modules instead of copy-pasting.
Local rollback snapshots for anything the platform will not read back, prod-mirror QA lattices, and singleton/throttle guards against race conditions under load.
Available for senior Genesys Cloud engineering, with a focus on AI and automation.