Evidence reviews
Structured assessment of claims, source quality, competing accounts, evidentiary gaps, and material uncertainty.
Research & AI systems
What I Do
I conduct independent, source-driven research and design AI systems and workflows for complex work. My focus includes public policy, law, government institutions, and administrative reform, alongside practical uses of AI that help people research, build, organize, and communicate with greater reach and clarity. I am available for selected paid engagements.
Open to selected engagements
Research I can provide
Legislative and regulatory research
Research into statutes, regulations, proposed rules, legislative history, institutional authority, and existing-law fit.
Public-record and public-source research
Focused collection and analysis of governmental, legal, institutional, and other publicly available records.
Issue mapping and policy comparison
Clear organization of complex questions, stakeholder positions, comparable approaches, tradeoffs, and implementation choices.
Source verification
Checking citations, tracing claims to authoritative materials, reconciling conflicts, and identifying unsupported conclusions.
Research briefs
Concise, source-transparent written products designed around a defined question, audience, and decision need.
Research and general legal analysis are provided for informational and policy purposes, not as legal advice or legal representation.
AI systems and workflows
Powerful tools, thoughtfully directed.
Good AI work begins before a model is asked to produce anything. I translate a real problem into a system of roles, context, tools, decision points, verification, and human review—then refine the arrangement until it is useful, understandable, and proportionate to the work.
The aim is not automation for its own sake. It is to extend what a person or organization can do while preserving the judgment, responsibility, and understanding that make the result worth using.
Workflow and system design
Turning complex or recurring work into a clear operating structure: what AI should do, what conventional software should do, where human judgment belongs, and how the pieces work together.
Research and knowledge systems
Designing source-aware workflows for discovery, triage, analysis, drafting, citation, revision, publication, and the preservation of useful context over time.
Context and instruction architecture
Creating the questions, instructions, roles, memory, and bounded work packages that let models operate with purpose rather than improvise from a thin prompt.
Review and quality control
Building challenge, verification, testing, uncertainty, correction, and approval into the workflow so fluent output is never mistaken for reliable work.
Human-centered automation
Combining deterministic processes with model judgment where each is strongest, while keeping the system understandable, stoppable, and accountable to its owner.
Working prototypes and tools
Developing practical dashboards, research systems, publication workflows, and local tools that turn an idea into something that can be used, evaluated, and improved.
If your business or organization is exploring where AI could genuinely help—or trying to turn scattered experimentation into a coherent way of working—describe the problem and what a useful result would look like.
Discuss an AI system or workflowA note on AI slop
It’s about supervision
AI is an extraordinarily powerful tool in the right hands. It can extend a person’s reach, reveal new ways to organize a problem, and make work possible that would once have required a much larger team. But capability is not the same as judgment.
“AI slop” is usually not the product of too much power; it is the product of that power being poorly directed. Thin prompts, absent context, unexamined output, and no meaningful standard of review produce work that is generic, unreliable, or merely abundant. Skilled use requires framing the problem, assigning the right role, supplying the right materials, supervising the process, and knowing when to revise, reject, or begin again.
I describe ChatGPT not as my assistant that does the work for me, but as my staff: capable of taking on different roles and extending what I can do, but requiring direction, division of responsibility, oversight, challenge, and final review. The questions, architecture, standards, and substantive decisions remain mine, and I remain responsible for the result.
The point is not to remove the human from the work. It is to give human judgment more reach.
Read about ARRP’s approachHow I work
Following the evidence.
Every project begins with a question and a willingness to follow the evidence wherever it leads. The work is shaped as it unfolds—bringing order to complexity, testing ideas against competing views, and revising both the argument and its form when the record demands it. The aim is not merely to arrive at an answer, but to make difficult subjects clearer and create work worthy of the public’s trust.
Research Method in Practice
ARRP
My principal independent project is the American Restoration and Resilience Project (ARRP), a sustained examination of structural problems in American government and potential institutional remedies.
Its published work demonstrates my approach to primary-source research, legal and policy analysis, methodological transparency, auditing, and revision.
Explore ARRPDOJ-001
Recent Presidential Personal Counsel in Senior DOJ Leadership
A primary-source and existing-law analysis of institutional conflicts, eligibility rules, and a proposed statutory cooling-off period.
Read the example ↗DOJ-003Politically Selective Enforcement, Charging, and Favoritism
An issue map connecting public records, charging comparators, statutory reporting structures, oversight design, and stated uncertainty.
Read the example ↗ELEC-011Algorithmic Redistricting Baseline and Representation Safeguards
A comparative policy and legal analysis developed into a model-state approach with auditable safeguards and explicit implementation limits.
Read the example ↗Project dashboard
Current and planned work.
A deliberately concise view of the work: research that is public or planned, alongside AI systems and tools being developed.
Loading current projects…
Research principles
Research designed to remain inspectable.
- Independence
- Engagements are accepted subject to research independence, transparent methods, and the protection of audience trust.
- Source transparency
- Material claims are connected to sources, with priority given to primary legal, governmental, and authoritative records.
- Uncertainty
- Gaps, conflicting evidence, assumptions, and unresolved questions are identified rather than concealed.
- Verification and accountability
- Drafts are checked, challenged, revised, and corrected. I remain responsible for the final research and analysis.
Work with me
Bring a defined question or problem.
If you need a source-driven review, issue map, policy comparison, research brief, or a thoughtfully designed AI system or workflow, use the contact form to describe the question, the intended users, the available materials, and what a useful result would look like.
Discuss an engagement