Procys | Production document AI and identity verification by Kapil ChauhanProcys | Production document AI and identity verification by Kapil Chauhan

Procys | Production document AI and identity verification

Kapil Chauhan

Kapil Chauhan

The problem Invoices and identity documents arrive in varied formats. Extracted fields need validation and a review path before they reach business systems.
My role As Lead AI Engineer at Procys, I own model selection, agent orchestration, evaluation, deployment and monitoring for the production AI platform.
The approach The document pipeline combines OCR with LLM reasoning about layout, then applies deterministic checks to header fields, line items and tax values. Confidence and comparisons with existing records route uncertain data to a reviewer with source context. Identity verification workflows use permissioned tools and auditable checks.
Delivered scope Structured extraction, validation, exception review and production operations across document processing and identity verification.
Like this project

Posted Sep 27, 2026

Led production AI systems that extract and validate document data, route exceptions for human review, and support auditable identity verification.