Data architect · Northern Virginia

I build systems that can
explain themselves.

I help regulated teams build data and AI systems that can be traced, defended, and trusted after the output leaves the screen.

See the problem Read the proof

Provenance

Lineage

Judgment

Governance

Art

Structure

Why this matters

AI made output cheap. Trust is still expensive.

A system can look finished until someone asks where the answer came from, what changed, and whether it can be defended later.

The question is no longer whether a system can produce an answer. It is whether that answer can be traced, reproduced, challenged, and defended.

That is where I do my best work: when AI, data, governance, and judgment have to hold together.

01 · Answers without evidence

The answer sounds right.

But no one can show which source, prompt, model state, retrieval path, or decision rule produced it.

01 · Answers without evidence

The answer sounds right.

But no one can show which source, prompt, model state, retrieval path, or decision rule produced it.

02 · Systems without memory

The dashboard changed.

But the lineage is thin, the handoffs are informal, and the record breaks the moment someone asks what moved.

03 · Automation without accountability

The risk arrives later.

After launch, after adoption, after an audit, after a client asks for proof. That is when the architecture has to answer.

When to bring me in

01Govern the pilot before it becomes production.AI pilots need governance before they become production systems.
02Add lineage before leaders rely on the metrics.Data platforms need lineage before leaders rely on the metrics.
03Set the standard before speed becomes institutional risk.Teams need a standard before speed becomes institutional risk.

The Trace Standard

I do not trust output I cannot trace.

Before I trust a system, I look for four things. Not a screenshot. Not a promise. A record another person can inspect.

01 · Source

Where did it come from?

The evidence stays attached to the answer instead of disappearing behind the interface.

02 · Reproduce

Can it be recreated?

Another person can follow the same path and understand how the result was produced.

03 · Change

What changed?

The record shows when data, models, prompts, or rules moved after the system shipped.

04 · Uncertainty

When should it stop?

Low confidence is visible, and the system knows when a person needs to step in.

Provenance trace

Source to answer, with the record intact.

Each answer keeps its cited source, lineage, confidence decision, and audit record attached so another person can inspect the path later.

Confidence

0.91

Hash

SHA-256

Action

Review

Live provenance trace

01 / 04

Trace sample · review path

Policy passage

The answer starts with cited source material, not a generated guess.

Confidence

0.91

Hash

SHA-256

Action

Review

The receipt

I built a retrieval system on federal policy documents to test the standard in working code. Every answer keeps its evidence, decision context, and audit trail.

Tests

Indexed passages

9,116

Citations, confidence, model context, and source hashes stay with every answer.

Inspect the system

Record 01

automated tests behind the retrieval pipeline

Tests show the retrieval path is being checked before trust moves to the answer.

Record 02

policy passages indexed with lineage

9,116

Lineage stays with the source material, so a reviewer can inspect the record behind the response.

Record 03

audit fields recorded for every answer

The answer record captures confidence, sources, model context, and review status instead of leaving judgment invisible.

Record 04

content hashing on every indexed record

SHA-256

Hashing makes source changes detectable, so the system can show when the underlying material no longer matches.

Receipts from the work

The standard matters because the outcome does.

Field record

Federal data pipeline

The friction

Replication took hours, and avoidable corrections kept pulling people back into the process.

The decision

I rebuilt the path across six sources with cleaning, deduplication, classification, and indexing inside the flow.

The outcome

99.98% accuracy · 40% fewer manual fixes · hours reduced to minutes

Field record

Governed data platform

The friction

Audit questions were slow to answer, storage was expensive, and transformation logic was hard to inspect.

The decision

I built a medallion architecture with visible ingestion, tested transformations, and a clearer record from source to use.

The outcome

50% faster audit queries · 30% lower storage cost

Field record

Audit-ready retrieval

The friction

A convincing answer was not enough. The system needed to show its evidence and expose uncertainty.

The decision

I built citations, confidence checks, model context, source hashing, and a query-level audit record into the answer path.

The outcome

44 tests · 9,116 passages · 16 audit fields · SHA-256

Two houses · One source

One standard, expressed through systems and objects.

The technical practice

Archetype Core

Data and AI systems built for inspection: traceable, defensible, and ready to stand up after launch.

Enter the house

The technical practice

Archetype Core

Data and AI systems built for inspection: traceable, defensible, and ready to stand up after launch.

Enter the house

The art house

The Ferne

Original work made with restraint, texture, and private meaning. Objects meant to be lived with and kept.

Enter the house

Still life of folded linen, ceramic vessel, and a sheet of paper in soft daylight

The origin

The work that made the standard.

For nearly six years, I built inside federal data environments tied to immigration cases, benefits processing, congressional inquiries, and records that affected real people.

That work taught me that almost right is not enough. Handoffs matter. Sources matter. The record underneath the visible system decides whether anyone can trust it later.

I founded Archetype Core to bring that discipline into the AI and data systems organizations are building now. The Ferne carries the same attention into material, memory, and objects meant to last.

Read the story

Thinking in public

I show the work while it is still becoming.

The writing carries the argument. The videos carry the build. The repository carries the proof.

Newsletter

Can your system answer for itself?

Mark what is true today. Not what is planned. Not what the vendor says the platform can do.

Current record

0/4

Start with one answer your team cannot fully explain. That is usually where the real architecture work begins.

Bring me an answer you cannot defend.

Review the system

I build systems that can
explain themselves.

AI made output cheap. Trust is still expensive.

The answer sounds right.

The answer sounds right.

The dashboard changed.

The risk arrives later.

I do not trust output I cannot trace.

Where did it come from?

Can it be recreated?

What changed?

When should it stop?

Source to answer, with the record intact.

Policy passage

The standard matters because the outcome does.

Federal data pipeline

Governed data platform

Audit-ready retrieval

Archetype Core

Archetype Core

The Ferne

The work that made the standard.

I show the work while it is still becoming.

Beyond the Prompt

YouTube

Audit-ready RAG

Can your system answer for itself?

I build systems that can explain themselves.

AI made output cheap. Trust is still expensive.

The answer sounds right.

The answer sounds right.

The dashboard changed.

The risk arrives later.

I do not trust output I cannot trace.

Where did it come from?

Can it be recreated?

What changed?

When should it stop?

Source to answer, with the record intact.

Policy passage

The standard matters because the outcome does.

Federal data pipeline

Governed data platform

Audit-ready retrieval

Archetype Core

Archetype Core

The Ferne

The work that made the standard.

I show the work while it is still becoming.

Beyond the Prompt

YouTube

Audit-ready RAG

Can your system answer for itself?

I build systems that can
explain themselves.