"The right metric for content quality is time-to-understanding."
Most people pick a lane: writer, manager, or engineer. I've never found a good reason to.
I've owned documentation at AWS and Pachyderm: writing for SageMaker AI, leading the DynamoDB docs team, and documenting MLOps infrastructure at Pachyderm. My background is in software engineering and technical program management.
I also built the infrastructure underneath: automated pipelines, agentic content systems, and tooling that speeds content production and helps keep documentation accurate at the pace of the product (that one is a hard problem whose solution I keep chasing), freeing writers for higher-judgment work.
This is a portfolio of some of my documentation and content systems work built from test accounts, design docs, and a working understanding of each system. Know your product, know your customer.
I care about this work more than is probably necessary. The Philosophy page goes deeper if you're curious. Let's talk.
SageMaker Training Plans — Full Feature Documentation
Greenfield documentation for a new GPU capacity reservation feature. API shape was changed based on documentation feedback. Full authorial control from concept to launch at re:Invent 2024.
DynamoDB Throttling Diagnostic Framework
Built from scratch on a new team. The four-category taxonomy and reason→metric reference table are deliberate information architecture decisions. First major deliverable on the DynamoDB team.
Pachyderm Production Deployment System
Complete deployment documentation system built and maintained as sole technical writer: cloud-agnostic architecture diagram, infrastructure and ingress layer, full production deployment guides for AWS, GCP, and Azure, and updated documentation for a proxy architecture change. The architecture diagram was carried forward into HPE's official product documentation after the acquisition.
EMR + SageMaker Integration — Rebuilt from Stale to Production-Grade
Stale documentation had triggered a prospect escalation. PM and GM requested an urgent update. Went deep independently into IAM role chaining, trust policy construction, and cross-account topology — and delivered a complete rewrite across same-account and cross-account paths, with limited SME support and a hard deadline. The prospect acknowledged the improvement.