The premise is fuzzy
You have an attractive idea, but the real user problem and the smallest honest test are still unclear.
Research strategy for high-stakes ideas
Turn uncertainty into a decision.
I help teams find the real problem, design responsible human–AI workflows, and build the evidence needed to move — before a promising idea becomes an expensive assumption.
Decision trace / 01
live question01What are we assuming?
02What would change our mind?
03What is the smallest useful test?
Clearclaim
Boundedrisk
Usefulevidence
confidence follows evidence
Background
PhD cognitive psychologist
Recent proof
Led a NIST-funded challenge winner
Working style
Direct, bounded, evidence-first
A useful place to start
You do not need more certainty theater.
The AI claim is ahead of the evidence
The workflow needs failure boundaries, meaningful human control, and a way to show what the system knows.
The pilot has to teach you something
You need outcomes, decision rules, and a credible story before a prototype becomes a costly commitment.
Ways to work together
Three bounded ways to get unstuck.
01
Problem and Product Discovery Sprint
A team with a promising idea that is not yet sure it solves the right problem.
The question
Whose problem is this, what do they actually do today, and what is the smallest test that would tell us whether we are right?
02Core focus
AI Evaluation and Human-Oversight Design
A team with an AI workflow that needs evidence, failure boundaries, and meaningful human control.
The question
What is this system claiming, how would we know if it were wrong, and where does a person need to stay in the loop?
03
Research-to-Pilot Strategy
A researcher, nonprofit, or early product team moving from an interesting concept to responsible field testing.
The question
What outcome are we changing, how would we measure it honestly, and what result would make us stop?
The working loop
Evidence should change the plan.
Research is only useful when it sharpens a decision. Every engagement ends in an action, a boundary, or a reason to stop — not a deck that politely restates the question.
- 01
Name the decision
Get specific about what must be chosen, by whom, and under what uncertainty.
- 02
Expose assumptions
Separate what is observed, inferred, hoped for, and still unknown.
- 03
Design the test
Find the smallest useful piece of evidence that could genuinely change the plan.
- 04
Make the call
Turn the result into a clear recommendation, including what would make it wrong.
Selected work
The proof includes the limits.
01
2025
Proposal and methodology
TwinSight
Winning entry in a NIST-funded national challenge on turning smart-city digital twin data into decisions people can act on.
02
2011–2023
Completed earlier in my career
Memory research, teaching, and mentorship
Peer-reviewed work on how acute stress shapes memory, alongside university teaching and graduate research mentorship.
03
2025–present
Working prototype
Assistant to the Caregiver
An early-stage Android application testing whether low-friction capture and caregiver-reviewed AI can reduce the bookkeeping burden between medical appointments.
complicated → clearWhy me
I study where people misread evidence — then design the next move.
I spent a decade researching human memory and teaching people to reason from evidence. Now I apply that discipline to product questions, responsible AI, pilot design, and the messy space between a persuasive idea and a defensible decision.
My strongest work starts where the answer is not obvious — and where pretending it is would be costly.