Meglio Alan

AI Engineer & Full Stack Developer

I build AI solutions that work in production, not just prototypes.

View projects Download CV

Building from Argentina

Projects

Projects in production and in construction.

01 — In production

AgendaGo

Multi-industry appointment-booking SaaS.

Three AI agents with tool calling:

Dashboard assistant with 21 tools that configures the system and proposes changes the owner approves.
WhatsApp agent that books appointments (app approved by Meta).
Supplier invoice and delivery note reading with vision.

Evals with in-house datasets (100 phrases; 52 scenarios and 15 WhatsApp conversations) and cost and latency telemetry per call.

React
TypeScript
Vite
Tailwind
shadcn/ui
Supabase (PostgreSQL, RLS, Edge Functions)
OpenAI
WhatsApp Cloud API
MercadoPago

See the productagendago.com.ar See the case study

02 — In production

Bonuxo

Multi-tenant loyalty platform: points, stamps, rewards, prize wheel, statistics, cards in Apple and Google Wallet.

Public API for POS systems, e-commerce integrations, recurring billing with MercadoPago.

Nux — In construction

In-panel assistant with an AI agent: tool calling with 19 tools, human confirmation on writes, an interchangeable OpenAI/Anthropic provider and per-plan cost caps. Integration in progress: LangGraph, LangChain, RAG and Langfuse.

.NET
PostgreSQL
Redis
React
Docker
Coolify

See the productbonuxo.com See the case study

Architecture diagram of the agents platform: data sources, ingestion service, PostgreSQL with pgvector, agent service with LangGraph and LangChain, and a web app with Next.js and FastAPI.

View full-size diagram

03 — In construction · phase 0: architecture

AI agents platform for companies

Connects Excel, databases, APIs and documents, and answers in natural language showing where each figure comes from.

Multi-agent orchestration with LangGraph, read-only SQL, RAG with citations, per-company isolation. Documented with a C4 model, a sequence diagram and 10 ADRs.

Python
LangGraph
LangChain
FastAPI
PostgreSQL + pgvector
Langfuse
Next.js
Docker

See the case study

Where I applied them

AI Engineering competencies

In production

AgendaGo and Bonuxo

AI agents and tool calling
Models that choose which tool to call, and with which parameters, to operate a real system.
LLM evaluation
Own datasets (100 phrases; 52 scenarios and 15 WhatsApp conversations) and per-call cost and latency telemetry.
Vision with multimodal models
Reading supplier invoices and delivery notes with multimodal models.
Integrating LLMs with business systems
Connecting models to the WhatsApp Cloud API, MercadoPago and POS APIs.
Observability
Per-call usage logging.
MCP (Model Context Protocol)
Own MCP server.

In construction

AI agents platform for companies and Nux (Bonuxo)

AI agents and tool calling
Nux: tool calling with 19 tools and human confirmation on writes.
Multi-agent orchestration with LangGraph
Agent graphs with a router, SQL and RAG subgraphs, bounded retries and memory with a checkpointer.
LangChain
Interchangeable models, tools, loaders, splitters and PGVector integration.
RAG and semantic search
Embeddings, chunking and pgvector to answer with a document and page citation.
Safe text-to-SQL
Natural-language queries over databases: validated SELECT only, a view allowlist, a read-only user and per-tenant RLS.
AI guardrails and safety
Input and output filters, prompt injection defense and multi-tenant isolation.
Observability
Traces with Langfuse.