🟢 I built an AI agent that reads invoices so humans don't have to. Manual invoice entry is slow, boring, and full of typos. So I automated it end to end in n8n.⚫ Here's how it works: 🟢 Invoice comes in: A webhook receives the PDF through a POST request ⚫ Text extraction: The file is parsed and the text pulled out 🟢 Safety check: If the file is invalid, the system responds with an error right away ⚫ Smart chunking: Large invoices are split into smaller pieces so nothing gets cut off or lost 🟢 AI extraction: A loop sends each chunk to an Information Extractor powered by Anthropics Claude, which pulls out the fields the client needs (vendor, invoice number, dates, line items, totals) ⚫ Merge and validate: The results are combined into one record and checked before they are sent 🟢 Clean JSON response: The structured data goes straight back through the webhook The hardest part was'nt the AI. It was making it production-ready. 💡 Long, multi-page invoices break most simple AI workflows because they exceed the models limits. Chunking, looping, merging, and validating are what let this handle large invoices reliably. Where this fits in the real world: ⚫ Accounting and finance teams processing hundreds of vendor invoices 🟢 Businesses syncing invoice data into ERPs or spreadsheets ⚫ Freelancers and agencies automating their billing workflows What used to take minutes per invoice now takes seconds, with no copy-pasting. ⚡ I'm currently going deeper into n8n and AI automation. If you are looking to automate document processing or any repetitive workflow, lets connect! 🤝 #n8n #Automation #AIAutomation #InvoiceProcessing #WorkflowAutomation #Claude #FullStackDeveloper #NoCode

I build AI that’s measured before it ships. Now open for freelance work.

Most AI features fail quietly. No one measured them, and the numbers come straight from the model. My work fixes both.

What I’ve built:

→ A multi-agent tax copilot for UAE–US businesses, live in production. The LLM reads the documents and cites the rule, and a deterministic engine does every calculation. 89% answer correctness on a human-reviewed eval.

→ An evaluation harness for legal AI. It benchmarked three frontier models on contract clauses (F1 0.88–0.90) and found a failure mode all three share: they cut off the legal condition that actually matters.

→ A brain MRI research platform with 3D lesion segmentation, sealed audit trails and FHIR export.

→ A pricing tool for metals traders that turns offer sheets (Excel, PDF, photos, voice) into priced bids.

→ Voice agents: a voice-to-clinical-note scribe, and a browser agent you control by speaking.

How I can help: • AI agents and RAG systems, with evals built in • Reliability audits for an AI feature you already shipped • Document extraction from PDFs, sheets and scans • Voice agents for calls and intake • Healthcare AI workflows

If you have an AI feature that’s unreliable, or one you haven’t built yet, send me a message. First 3 clients get a priority start this month.

#AIEngineering #LLM #AIAgents #RAG #HealthcareAI

89% correctness on a human-reviewed eval for the tax copilot is a solid bar, good luck with freelance!