Professional Portfolio

Sabrina Picari

AI Workflow Automation Specialist

I design and build advanced automation pipelines on n8n, integrating AI Agents (OpenAI LLM), App integration (API), and cloud platforms (SharePoint). I bridge the gap between technical systems and editorial operations to scale content production, automate complex web scraping, match catalogs, and coordinate multi-language translations.

LinkedIn Profile

Workflow Automation

Low-code n8n integration with decision-making AI agents.

n8n OpenAI API

Automated n8n Workflows

In Development (WIP)

1. Web Review Scraping & Catalog Match Engine

A complex pipeline designed to scan external sitemaps, crawl product review pages, extract metadata using a hybrid parser (JSON-LD + CSS + AI fallback), query catalog databases via OpenSearch fuzzy queries, and semantically verify results using AI agents to prevent catalog contamination.

Scraping & Analysis Time ~5 Minutes vs 1 Day manual
Automated Translation (WIP) -2 Hours per execution
Explore Technical Details
Active in Production

2. Product Test PDF Extraction & Multilingual Distribution

An event-driven automated pipeline that parses technical product test PDFs from SharePoint. It extracts test metrics and ratings using OpenAI visual parsing agents, queries OpenSearch to match catalog items, localizes the verdict (Fazit) into 4 languages in parallel, and uploads localized CSV feeds back to SharePoint target folders.

Processing Time A Few Minutes vs 2 Hours manual
Scheduled Execution 100% Automated Weekly run
Explore Technical Details
AI DE

Technical Expertise & Toolkit

n8n Workflow Automation

  • Designing complex, multi-branch workflow architectures
  • Managing batch iteration loops and conditional routing
  • Implementing error handling, retries, and alerts
  • Advanced REST API integration and custom HTTP nodes
  • Event-driven orchestration and scheduled cron triggers

AI & LLM Integration

  • Configuring structured decision-making OpenAI agents in n8n
  • Prompt Engineering for guaranteed output validation
  • Semantic matching algorithms to resolve catalog entries
  • Automated editorial translation and content localization
  • Context compression to minimize token counts and API costs

Data & Search Operations

  • Executing fuzzy queries on OpenSearch databases
  • Data manipulation and sanitization using custom JavaScript
  • Parsing unstructured HTML, JSON-LD, XML, and PDF streams
  • Bi-directional sync with cloud storage systems (SharePoint)
  • Custom CSV generation for localized publishing feeds