Work

Professional

Enterprise Context-Aware Conversational AI Platform

Enterprise conversational AI gateway grounding LLM responses on real business entities — clients, projects and active persons — from an entity catalog.

Overview

The platform is an enterprise AI system powering AI-assisted conversations, meeting intelligence, and background job processing for business workflows across three microservices.

What I Built

Chat API refactor (Core Conversational AI Gateway)

  • Refactored the core chat service and DTOs into a modular structure with improved context handling and response generation.
  • Wired Entity Catalog data — clients, projects, active persons — directly into chat context and LLM prompt construction, so responses are grounded on real business data instead of generic completions.
  • Fixed the project query to filter by active person status in the join conditions.

Global response standardisation

  • Added a response-envelope opt-out for endpoints that need raw passthrough instead of the global response envelope.

Technology

NestJS · TypeScript 5 · Node.js 22+ · Clean Architecture · LLM prompt engineering · Google OAuth · JWT · Docker

Jump to