Nathan Payne

Nathan Payne

Most product decisions are invisible—until they aren’t.

Context

For ten years, I worked on the streaming platforms behind Disney+, Hulu, and ESPN. I started at MLB.com and BAMTECH, joined Disney through the 2017 acquisition, and spent five years leading the Native Client Platform across PlayStation, Xbox, Amazon Fire TV, Linux set-top boxes, and smart TVs.

Approach

I like working through problems where systems disagree, the rules leave something unresolved, or a product claims more than it can prove: a codec mismatch, an ambiguous financial rule, an AI agent that knows a policy but skips it. I work out which source to trust and what the product needs to do, then make that clear enough to design, test, and operate.

Now

I build and operate products with Claude Code, Codex, and Cursor. I also built Mergepath, the review and enforcement system those agents work under. That work keeps bringing me back to three questions: what evidence is enough, which decisions an agent can make, and when I need to decide. Open to senior product and platform roles, especially in streaming infrastructure, developer tools, and workflows that use AI.

Read my résumé
October 2026
Projects

Selected Projects

The case studies focus on the decisions, tradeoffs, and evidence behind them; I built each with AI agents under a review system I designed.

Multiplayer bingo I built in eight days for a cruise and ran ten days at sea. The standings held, but the social mechanics barely caught on. A second event showed reusing the engine wasn't enough.

A code-review standard for AI coding agents. Written rules didn't stop them pushing to main, so I built gates that could refuse. An audit then found eight of ten repos requiring none of five checks.

A Broadway deal model I built from my own investment documents, to learn the domain and test AI coding tools. It connects terms to investor outcomes but still hasn't run a real capitalization.

A device registry I built at Disney: five feeds, one schema. AI drafted the heaviest intake; every commit still needed an admin. The demo shows the workflow on invented data.

A career CRM that stops claims I can't trace to approved evidence. Paused before launch at 48% extraction and 19% match against an 80% bar I set before I'd built the ruler to read it with.

Swipe WatchEXPERIMENT

A swipe prototype that asks for taste before a viewer presses play. I let the demo promise better recommendations it didn't produce, so I couldn't separate reactions to the idea from the payoff.

Shared billing for nine people, with no recipient accounts. It took over the explaining and checking, but I kept getting the model right and stopping before the screen someone needed.

Consumer · Enterprise · Finance · Developer tooling
Community

Giving & Governing

Outside work, I’ve spent years organizing around causes I care about—raising more than $100,000, building campaigns, and staying involved after the fundraising ended.

  • Larkin Street Youth Services
    $16,684 raised
    Raised for Safe & Seen, a campaign for LGBTQ+ youth experiencing homelessness in San Francisco, supporting housing, education, and healthcare through Larkin Street Youth Services.
  • San Francisco AIDS Foundation
    $80,195 raised
    Three AIDS/LifeCycle rides: $9,166 in 2023, $45,646 in 2024, and $25,383 in 2025, supporting HIV prevention, care services, and housing stability through the San Francisco AIDS Foundation.
  • Jewish Pride Fund
    $8,959 raised
    After the campaign, I joined one of the country’s few LGBTQ+ Jewish giving circles to interview grant applicants and help decide where its grants go.
ScopeFundraising · Campaign Building · Grantmaking · Community Organizing
Connect

Connect

Get in touch · Book time · Résumé · Privacy

Elsewhere