Founding Engineer - Harness Optimization
Confirmed on the employer's own hiring board on Sep 26, 2026. First seen by Alion on Aug 22, 2026.
Company
Pax Historia is creating the future of interactive entertainment.
We’re building the platform for worldbuilders, storytellers, and game devs to create and publish any ‘what if’ they can think of. So far, most of the 25k+ user published presets on our site are historical (eg, ‘what if Rome never fell?’), but fantasy and sci fi are the fastest growing categories on the site.
We have seen over 92 million rounds played and raised a 10M+ seed round in March. We’re backed by Y Combinator (W26), Bessemer, Pace Capital, Z Fellows, and more.
Role
Each game on Pax Historia is run by the AI model the user picked to run their game. Games are much more fun if the AI running their game is able to balance long-term internal consistency with creative developments - in short, if the AI is good at storytelling.
But today, we’ve barely scratched the surface of getting the best possible results out of the ~240B tokens processed on our site each week. The same default AI workflows run across very different models, and we haven’t done much to adapt to the quirks of the 37+ models on our site now. We think our games can be much more interesting than they are today, which is why we’re hiring for this role.
On day one, you’d experiment with workflow architecture and prompts to begin making improvements and building the evals needed to figure out whether you’re actually making games better.
You might be a great fit if you’re extremely experiment-driven, a big-picture systems-thinker, and a storyteller yourself. We don’t expect you to have done this exact job before, but we are expecting you to be a fast learner with extremely strong fundamentals.
Logistics
You’ll be our 7th team member. We work in-person in San Francisco, 5+ days/week. We are open to sponsoring visas.
Comp will include meaningful equity to reflect your responsibility as a founding engineer.