EH/BVP

AI SAFETY / GOVERNANCE / COMMUNICATIONS / PRODUCT

I build practical tools for difficult questions about AI.

My work connects governance, communications, and product design. I turn open questions into interactive tools, programs, and clear public narratives.

Explore the flagship Run the live protocol
300KCommunity built
20M+One-week organic reach
FULL STACKGovernance prototype shipped
POLICY × PRODUCT × STORYCross-disciplinary synthesis

AI SAFETY / COURSEWORK, PROGRAMS, AND WRITING

What I have done.
What I am exploring now.

I have pursued structured coursework, competitive programs, and philosophical writing. My current essay asks when an AI system may understand something its overseer cannot.

01 / PUBLISHED WORK2 essays shipped

AI Philosophy Competition

Two essays completed through Zachary Goodsell's AI Philosophy Competition.

02 / COURSEWORKCertificate earned

BlueDot AGI Strategy

Completed BlueDot Impact's AGI Strategy Course and received a certificate.

03 / PROGRAMAccepted

AISC

Accepted into AISC, extending my engagement with structured AI-safety work and community.

04 / ESSAY IN PROGRESSTHE DECODER GAP

When the actor can read more than the overseer.

“Risk tracks the decoder gap: what the acting model can decode minus what its overseer (safety training, a guard model, or a monitor) can decode.”

Developing thesis, paraphrased for this portfolio.

FLAGSHIP PROJECT / PROOF OF EXECUTION

01 / SHIPPED AND LIVE

Respect Under Uncertainty

Respect Under Uncertainty is an interactive governance project about potential machine self-awareness. It asks people to make a decision without treating uncertain evidence as proof or dismissal.

01 / THE GAP

Institutions have to act before the evidence is settled.

Potential machine experience creates a three-sided governance failure: assume too readily, dismiss too quickly, or remain unprepared while uncertainty persists.

02 / THE BUILD

The prototype turns a judgment into a protocol.

Five fictional cases ask people to separate evidence, incentives, human risk, and possible welfare. A constrained model then pressure-tests the decision without acting as an oracle.

03 / THE NEXT TEST

Now it needs expert review and real pilots.

The next stage is expert review, facilitated pilots, stronger evaluation, and an open toolkit that safety practitioners and educators can adapt.

Test the theory Open the governance lab

HOW I WORK / FOUR PRACTICAL MODES

02 / GENERALIST RANGE

I research, build,
convene, and communicate.

01

STRATEGY / SCOPING

Choose a useful question

I look for important questions that are not yet supported by practical tools. Then I define a project small enough to test.

02

PRODUCT / OPERATIONS

Build the first version

I move from a written idea to an interface, pilot, or program. I test it with real users and revise it.

03

PROGRAMS / CONVENING

Work across disciplines

I bring together researchers, practitioners, writers, and designers when the problem needs more than one kind of expertise.

04

COMMUNICATIONS / TRANSLATION

Explain it clearly

My communications background helps me translate complex work without pretending the uncertainty has disappeared.

TRAJECTORY / SKILLS THAT LED HERE

03 / SIGNALS CONVERGE

My earlier work led here.
Each step added a useful skill.

Political science taught me to study institutions. Audience-building taught me how ideas move. Speculative writing taught me to make future choices concrete. I now apply those skills to AI safety.

  1. 01

    POLITICAL SCIENCE

    Studied how institutions distribute power and govern what they cannot fully predict.

  2. 02

    AUDIENCE SYSTEMS

    Built a 300K-person media community and learned how trust, framing, and incentives move attention.

  3. 03

    SPECULATIVE WORLDS

    Developed fiction and visual work about AI, climate, identity, and institutional choice.

  4. 04

    AI GOVERNANCE

    Shipped an interactive lab that turns moral uncertainty into a concrete decision process.

THE FULL CREATIVE + TECHNICAL RECORDEnter the Time Capsule

NEXT BUILD / 12-WEEK DIRECTION

The next version needs experts, pilots, and stronger evaluation.

  1. WEEKS 01-02Scope

    Map the highest-value users, expert objections, and adjacent work.

  2. WEEKS 03-07Build

    Refine the cases, evaluation layer, and facilitation format.

  3. WEEKS 08-10Pilot

    Run structured sessions with practitioners and emerging governance talent.

  4. WEEKS 11-12Ship

    Release the open toolkit, policy brief, facilitation kit, and findings.

BEST VERSION PROJECT / OPEN TO COLLABORATION

I want to build useful AI-safety projects with other people.

I am looking for collaborators and teams working on problems that require research, project design, clear communication, and steady execution.

See what I shipped Explore the full archive