Reworked an existing dashboard from a VAPI/n8n setup into an ElevenLabs batch-calling console for International Property Alerts. Users pick an agent, upload per-timezone lead lists via CSV, then fire and schedule call batches. Per-call outcomes sync back onto leads using the agent's disposition analysis, with global lead search, recall flows, timezone-aware scheduling, lead edit history via Postgres triggers, and links to the ElevenLabs conversation recordings.

I then added two-way AI texting. An AI persona named "Jack" texts leads after three unsuccessful calls, holds a natural SMS conversation to book appointments, and auto-extracts lead state each turn (interested, opt-out, or appointment time) back into the dashboard. It includes YES/STOP consent compliance, a human take-over mode, a unified per-lead call and SMS timeline, and call-log backfill, kept fully isolated from the adjacent production system in the shared database by design.

• RoleFull-stack & AI engineering
• PlatformWeb Application
• StackNext.js, Supabase (Postgres, RLS, triggers), ElevenLabs Conversational AI, Telnyx Messaging (10DLC), Anthropic Claude (Sonnet), n8n, Vercel