Skip to content
John Charlie CatedrillaSelected work

Representative work

Document Workflow Systems

Software for document-centric processes: ingestion, extraction, human review, and status tracking, built to stay auditable when the input is messy.

This is a representative work note, not a public product case study. It describes the class of problem, the scope of the work, and the engineering practices involved; client identities, product specifics, figures, and implementation details are deliberately withheld to respect confidentiality.

The problem

Organizations run on documents — intake, review, extraction, approval — and manual handling is slow and error-prone. Each document type brings its own formats and edge cases, so the system has to normalize messy input without losing the trail of who changed what.

Scope of the work

  • Build document-centric workflow features end to end: ingestion and parsing, data extraction, human review steps, and status tracking.
  • Model workflow state so a document's stage, assignee, and history are queryable rather than implied by folder structure or memory.
  • Implement background processing for parsing and extraction jobs, with retries that behave predictably after partial failures.
  • Work within employer-owned products and existing team processes; the workflows themselves belong to the organizations that run them.

Engineering considerations

  • Separate extraction from verification: route low-confidence results to a human instead of shipping silent errors.
  • Validate and normalize input at the boundary so downstream steps inherit clean data, not cleanup work.
  • Make processing idempotent; a retried job must not duplicate documents, records, or notifications.
  • Keep an attributable trail of changes to documents and extracted values — audit questions arrive after the feature ships.
  • Design review screens around the reviewer's actual sequence of decisions, not the shape of the data model.

Capabilities demonstrated

  • Document parsing and extraction pipelines
  • Human-in-the-loop review design
  • Workflow and state modeling
  • Background job and queue handling
  • Full-stack feature delivery

Technology categories

  • Python and TypeScript services
  • Node.js APIs
  • PostgreSQL data models
  • Workflow automation tooling
  • React and Next.js interfaces