Anthony Béal

Technical takeover of a gaming optimization desktop application (stack: Tauri v2, SvelteKit, Next.js 15, Express v5, Prisma v6, PostgreSQL 16, C++/C#). Mission started with a complete technical audit of the existing monorepo (architecture, technical debt, blocking points), followed by the deployment of Phase 1: beta version release. Bug fixes and new feature development.

Development and maintenance of a digital signage content management SaaS platform (Digital Out-Of-Home).

Recurring fixed-price intervention on the design and evolution of the Flutter mobile application, implementation of a multi-tenant architecture with Supabase (RLS, Auth, Edge Functions TypeScript), and development of content publishing workflows.

Stack: Flutter, Supabase (PostgreSQL, Auth, Edge Functions), TypeScript, CI/CD.

Complete development of an automation system to manage travel requests via WhatsApp, for a travel agency.

The mission's objective was to automate the first phase of customer qualification while maintaining a natural and personalized experience.

I designed and implemented an intelligent conversational bot based on LangChain and LangGraph, capable of analyzing prospect messages, detecting dates, destinations, and intentions, and then asking relevant questions to build an actionable travel file.

The bot operates on the WhatsApp Business Cloud API, with full webhook management, message reception and sending, template validation, and status tracking.

On the technical side, I developed the entire backend: server architecture, endpoints, extraction logic, conversation context management, and saving necessary information for the smooth running of the flow. The automation relies on robust state logic, allowing the bot to know where each customer is in their journey.

A detailed logging system was added to track interactions, facilitate monitoring, and enable quick adjustments in production.

Everything is deployed on a Linux environment, with stable and supervised configuration, ensuring the bot's reliable daily operation.

This mission enabled the client to automate a large part of the initial request processing, improve responsiveness, and save time while maintaining a professional and human tone.