Data Pipeline Orchestration with Dagster

Context

The growth of analytical infrastructure and the need to integrate data from multiple complex transactional sources (such as ERPs on SAP HANA and storage and logistics systems on SQL Server / Intralog) required a centralized, auditable, and resilient system.

To solve this, the old model of cron tasks and isolated scripts was migrated to a data architecture fully orchestrated with Dagster at Vinos América, structured through independent gRPC microservices coordinated by Docker containers.

AI Integrations with Model Context Protocol (MCP)

Context

To boost decision-making and empower advanced AI assistants (like Claude or ChatGPT) with the ability to query and analyze information in real time, a microservices ecosystem based on the Model Context Protocol (MCP) was designed and implemented.

This ecosystem acts as a secure, governed, and high-performance bridge between Large Language Models (LLMs) and the corporate Data Warehouse (DWH) at Vinos América.

[!NOTE]
Confidentiality and Intellectual Property Note:
Being a private and strategic enterprise development, the source code, credentials, and the Data Warehouse (DWH) codebase/schemas are not publicly available (not open-source) to protect information security and organization operational rules.

VA Proyecciones: Demand Forecasting Statistical Library

Context

In the inventory planning and distribution workflows of Vinos América, there was a requirement to accurately automate the demand forecasting of thousands of combinations of product_id + warehouse_id.

To solve this challenge in a robust and industrialized way, VA_Proyecciones was designed and implemented, an internal modular Python library specializing in demand classification, statistical modeling, and backtesting of time series.

[!NOTE]
Confidentiality and Intellectual Property Note:
Being a predictive analytics library at the core of the company’s business, the source code and connection variables are not public (not open-source) in order to protect formulas and the organization’s competitive advantages in distribution.

Vinos América Data API: Database Integration and Custom GPTs

Context

Before the global development of the Model Context Protocol (MCP) ecosystem, there was a need to connect language models (specifically OpenAI’s Custom GPTs via Actions) quickly, securely, and directly with the company’s transactional system to enable real-time queries.

To solve this need, Vinos América Data API was designed and built, a robust REST interface that translates the AI’s natural language requests into secure, high-performance SQL queries against the SAP HANA enterprise database.