# AI Developer — 4 Document Automation Systems for CRE Firm ## The Engagement We're an AI consulting firm hiring one Python developer to build four document automation systems for a commercial real estate investment firm (50+ acquisitions/year, 700+ tenants, ~50 offering memorandums received daily). All four systems share Python, Claude API, n8n, and Microsoft Graph API — so one developer across all builds means major code reuse and efficiency. Total engagement: 10–14 weeks across four phased builds. ## The Four Systems **1. Lease Abstraction (Weeks 1–6):** Upload a 200+ page lease package → AI extracts rent, escalations, renewal options, obligations, and red flags → cross-references amendments to flag conflicts → outputs a structured summary. The hard part: tracking which provisions have been superseded by later amendments. **2. Environmental Report Processing (Weeks 4–8):** Upload a 150–250 page Phase I ESA → AI runs three extraction passes (findings, context, deal impact) → outputs a pass/flag/fail recommendation. Shares ~60–70% of code with System 1. **3. PSA Generation + Default Letter Automation (Weeks 6–9):** Two workflows — (a) generate formatted Purchase and Sale Agreements (.docx) from deal terms, and (b) read tenant data from Yardi daily, detect defaults/expirations, generate personalized letters, route through approval queue, send via M365. Zero false positives required. **4. OM Database (Weeks 9–12):** Continuous pipeline — n8n detects new OMs via email/SharePoint → AI extracts ~20 fields (cap rate, price, tenant, lease terms) → pushes to Airtable → daily digest email. Processes 50+ docs/day indefinitely. No custom frontend — Airtable is the UI. ## Required Skills - **AI/LLM APIs** — production apps using Claude or GPT-4 for document extraction AND generation - **Python** — production-quality; the entire engagement is Python - **PDF processing** — PyMuPDF, pdfplumber, or similar; OCR with Tesseract or Textract - **n8n or similar** — automated multi-step pipelines - **Microsoft Graph API** — Outlook email + SharePoint file integration - **python-docx** — programmatic Word document generation - **Cloud deployment** — AWS, Azure, or GCP **Strong plus:** Yardi/property management API experience, commercial real estate document familiarity, Airtable API ## What We Provide - Sample documents for all four systems (leases, ESAs, PSAs, OMs) - Microsoft 365 admin access via client IT - Yardi API docs and test environment - All templates and extraction schemas - Fast feedback — deliverables reviewed within 24 hours ## To Apply Answer all four questions — applications that skip will not be reviewed: 1. **Describe the most complex document extraction or generation system you've built using AI.** Tech stack, document types, accuracy achieved, how you handled inconsistent formats. 2. **Have you integrated with Yardi or a similar business platform (ERP, CRM, property management)?** If not, describe your most similar enterprise API integration. 3. **Scenario:** A lease signed in 2015 has 2% annual escalations. A 2019 amendment changed escalations to 10% every 5 years for years 6–10 only. A 2022 amendment extended the lease by 5 years but didn't mention escalations. What's the current structure, and how would you resolve this automatically? 4. **Link to relevant work** — GitHub, portfolio, or case study. Required. *Confidential engagement — do not reference the end client publicly.*