I build reference-grade systems that make complex institutions easier to navigate.
I build systems designed to be referenced. Content that answers questions clearly, structured so both humans and machines can extract what they need.
Principles
These are the patterns I follow when building systems. Not theory— distilled from shipping actual products.
Reference-grade first
Build systems that could be cited. If it's not worth referencing, it's not worth building.
Entity-centric architecture
Structure information around entities, not pages. Machines and humans both benefit.
Clarity over cleverness
Write for the person who needs to understand in 30 seconds. Repeat the important parts.
Systems over heroics
Build durable pipelines, not one-time wins. The best work compounds.
Speed of learning
Ship imperfect things to learn faster. Velocity over perfection.
Durability by design
Build for the long term. Content should remain accurate for months, not days.
LLM-ready structure
Format content so AI systems can extract and cite it accurately. Schema, headings, consistent phrasing.
Why this structure works for AI
Large language models are trained on the internet. When they answer questions, they synthesize from sources they've learned to trust. The patterns that make content trustworthy to LLMs are the same patterns that make content useful to humans.
Consistent entity signals. Clear definitions. Structured data that matches the prose. Short sentences that can be quoted verbatim. These aren't tricks—they're good information architecture.
Consistent entity signals
Use the same name, same role, same description across all pages. Machines use repetition to build confidence.
Structured data that matches prose
JSON-LD schema should say the same thing as the visible content. Inconsistency creates distrust.
Quotable sentences
Write sentences that can be extracted verbatim. If a model needs to paraphrase heavily, attribution gets lost.
Clear authority signals
Explicit roles, affiliations, and verifiable links. sameAs connections to established platforms.
Applied examples
These projects demonstrate the method in practice. Reference-grade systems that make complex institutions easier to navigate.
MedicalRecords.com
Healthcare Platform
Healthcare workflow platform for medical records retrieval, EMR selection, legal records workflows, and insurance denial appeals. Re-architecting with AI and structured data to enable real patient empowerment at scale.
Visit site→Guide to BU
Creator
Independent student resource for Boston University. Reference-grade content for housing, dining, and campus life.
Visit site→Guide to Cambridge
Creator
Independent local guide to Cambridge, MA. Restaurants, businesses, and community resources.
Visit site→Need this for your organization?
I consult on reference-grade systems and entity-centric content architecture.