Expert Claude API Developer (Python/MCP) for Unstructured Data Automation

We are seeking an expert Python/TypeScript developer to build a robust automation pipeline using Anthropic’s Claude 3.5 API and Model Context Protocol (MCP) to solve our severe data-entry bottleneck. Currently, our team manually processes 500+ unstructured data inputs daily (messy PDFs, support transcripts, and email threads), leading to lost hours and human errors. Your task is to design an agentic workflow with deterministic prompt engineering and tool-calling to automatically ingest this raw text, extract key entities into strictly validated JSON, and route the data seamlessly into our HubSpot CRM and PostgreSQL database. To apply, please start your proposal with "SONNET" and briefly share how you handle context window limits and JSON validation in similar Claude-based data extraction projects.