Why Chunking Breaks Business Documents
Chunking is not just a retrieval setting. In business documents, the wrong chunks can separate values from labels, tables from headers, clauses from definitions, and evidence from the answer.
Read ->Segment
Document AI
Projects, deep-dives, and implementation notes in this AI engineering domain.
Segment library
Projects and posts
Read this lane as a mini-curriculum: start with the project, then use the supporting posts to understand the engineering choices.
Why PDF Extraction Is Harder Than It Looks
PDF extraction fails in production because business documents are not clean text. They carry layout, tables, repeated fields, scans, conflicts, and evidence requirements.
Read ->PolicyTrace Conflict and Arbitration: When Source Documents Disagree
How PolicyTrace handles disagreement between source documents with deterministic authority rules, conflict entries, evidence-backed review, and reviewer-visible arbitration.
Read ->PolicyTrace Evidence and Provenance: Why JSON Is Not Enough
How PolicyTrace turns extracted insurance fields into reviewable evidence with hidden field citations, Docling geometry, provenance matching, PDF highlights, and reviewer-visible source context.
Read ->PolicyTrace Prompt Design: From PDF Text to Reviewable Evidence
How PolicyTrace designs prompts as one bounded layer in a reviewable Document AI system: document classification, specialist extraction, typed outputs, field citations, and evaluation gates.
Read ->Evaluating PolicyTrace: Golden Examples, Conflicts, and Review Outcomes
How I would evaluate PolicyTrace beyond simple accuracy: golden examples, conflict fixtures, provenance checks, review outcomes, regression gates, and operational metrics.
Read ->What I Would Improve Before Taking PolicyTrace to Production
A production hardening roadmap for PolicyTrace: authentication, durable storage, background jobs, audit trails, evaluation, monitoring, retention, and operational ownership.
Read ->Deploying PolicyTrace with GitHub, Docker, and Hugging Face Spaces
How PolicyTrace packages a FastAPI backend and React review UI into one Docker app for a Hugging Face Space, and what still separates a public demo from production deployment.
Read ->Designing the Human Review Loop for AI Extraction
How PolicyTrace turns extraction into a reviewer workflow with split-screen PDF evidence, field highlighting, confidence signals, verification, flags, and overrides.
Read ->The Golden Record Problem
How PolicyTrace turns overlapping insurance PDFs into a reviewable Golden Record by separating extraction from arbitration, conflict handling, evidence, and human review.
Read ->PolicyTrace
A Document AI workflow for UK motor insurance PDFs, with structured extraction, provenance, conflict resolution, and human review.
Read ->PolicyTrace layered architecture with evidence and review loops
A layered architecture map for PolicyTrace, showing where parsing, privacy, model calls, evidence artifacts, arbitration, and human review belong.
Read ->