Marcelo Vasconcelos
← Selected work

Chat App Crédito

A communications workspace for customer service, campaigns and AI-assisted conversation review.

TypeScript · Node.js · Fastify · Vue 3 · Pinia · PostgreSQL · Drizzle · Redis · BullMQ · WebSocket · Linux VPS · Caddy/TLS · Docker · GitHub Actions

Implemented communications application

Customer-service workspace · demonstration data
Customer-service workspace · demonstration data
Kanban workflow · demonstration data
Kanban workflow · demonstration data

Actual customer-service and Kanban interfaces rendered from the application build in an isolated, read-only preview. All contacts, messages and phone numbers shown are fictitious.

THE PROJECT

The problem

Customer service and sales teams need a shared view of conversations, channel activity and follow-up. Audio messages and distributed integrations add operational complexity.

My contribution

I implemented the communications application with AI-assisted tools, including the Vue interface, Fastify services, persistence, background workers and channel integrations.

Engineering decision

Channel delivery outcomes and uncertain provider responses are tracked explicitly to avoid blind retries. Local transcription is available for audio processing within the service environment.

IMPLEMENTATION

What I built

  • A Vue 3 customer-service interface and TypeScript/Fastify backend with Drizzle and PostgreSQL.
  • WhatsApp integration, encrypted session persistence and real-time updates over WebSocket.
  • Redis and BullMQ background jobs for campaign processing, with delivery reconciliation and retry safeguards.
  • Local faster-whisper audio transcription, AI-assisted conversation evaluation and reporting; voice and SMS integration workflows.
  • Linux VPS deployment using Caddy for reverse proxy and TLS, PostgreSQL/Redis containers, systemd-managed services and automated GitHub Actions deployment.

Outcome and current stage

A communications application with documented customer-service, campaign and AI review workflows. Provider configuration and activation requirements are kept explicit in the project documentation.

Private source repository. Screenshots and selected implementation details are shared without exposing customer records or private code.

Explore other projects →