Precision Pharma Logistics: Catalog Orchestration Engine

This project serves as a full-scale MLOps and LLMOps solution designed for an engineering request at Redcare Pharmacy in Cologne. It provides a robust pipeline for handling real-time catalog updates from multiple pharmaceutical vendors, ensuring data integrity through automated validation before reaching the modeling stage.

The system architecture integrates a suite of modern engineering tools, including Scikit-learn for predictive modeling and MLflow for lifecycle tracking and model registration. Orchestration is managed by Prefect to ensure a seamless flow from ingestion to serving via a containerized FastAPI. Additionally, the project incorporates advanced LLMOps practices by utilizing prompt-based demand classification, alongside Great Expectations for rigorous data quality checks.

To emulate a production environment, the entire stack is containerized with Docker and includes local MinIO storage to simulate AWS S3. A complete CI/CD workflow is implemented through GitHub Actions, covering linting, testing, and image deployment. This repository reflects a deep expertise in building scalable, reliable, and automated machine learning infrastructure for high-stakes business environments.