I'm Jeremy McKeehen, technical founder and product engineer. Co-founded MetaTravelers: motion comics, a sold-out digital launch, action figures in collectors' hands. I build the commerce and agent systems behind it, and I'm the merchant running them.
10,000+ physical units shipped · 22 countries · 2-person core
01 · proof terminal
Scroll to examine the systems underneath.
Smart contracts, frontend, generative 3D, automated rendering, cloud infrastructure, and delivery held through the public burst. The permanent audit trail is public.
7,777 minted · 2,689 unique buyers · 2,935 units in the public tranche · Ethereum · 0x5372f9…e056E5 · Dune source ↗
Ten thousand units to 22 countries with no order shipped in error. Between our Shopify storefront and our 3PL sits a layer that rebuilds each order, releases only what's shippable, and rejects anything ambiguous.
Reconciliation passes across preorders, wholesale, and DTC matched 893 shipments. Zero unidentified.
Smart contract, generative 3D pipeline, automated rendering, and delivery for the MetaTravelers digital launch.
Staggered the release to shed load off the public window, with caps enforced at the frontend and in the contract so an oversubscribed sale couldn't exceed supply.
Premium 7-inch action figures with up to 29 points of articulation — designed with Four Horsemen Studios, written by Andy Lanning — through factory, crowdfunding fulfillment, and Shopify as one physical goods pipeline.
Ranks the single human touch that would unblock the most valuable agent work, each with a plain-English reason it surfaced now.
Open source (MIT) on GitHub.
Verbatim zenoh keys broke liveliness matching for the /_schema path.
Windows tray declared browser capability only when a shared gateway token was present, leaving the UI stuck.
A failed SQLite close left the FD open while the byte-probe guard thought nothing was live.
Deep-link open of an instance skipped the connect path that refreshes module metadata.
Workbench omitted the WITHSUFFIXTRIE flag on RediSearch text field definitions.
AI is useful here because the work is real: product support, content pipelines, research, build logs, fulfillment notes, technical plans, and human approval around anything with stakes.
BEFORE: Research, production updates, support context, content planning, and build logs lived across scattered tools and founder memory.
AFTER: Agent workflows route work through an orchestrator, queue, executors, and human approval, so a two-person company ships, logs, and learns like a larger one.
I use Sidekick for theme Liquid, Page Builder blocks, and landing page variants for paid A/B tests. Most recently to audit the store for agentic sales, AEO, and GEO.
Looking for product engineering roles with wide scope: AI-first product teams, commerce platforms, and zero-to-one environments where ownership beats job descriptions.
