-
-
TYPE OF WORK
Full Time
-
-
WAGE / SALARY
$800/month
-
-
HOURS PER WEEK
TBD
-
-
DATE UPDATED
May 29, 2026
Remote · [Full-time / Contract-to-hire — TBD]
This is a brand-new role — the first dedicated hire for Bitrograde Labs. You won't slot into someone else's process; you'll help define how a small studio ships complete games fast by directing AI. Ground floor, real latitude, real ownership.
About Bitrograde Labs
Bitrograde Labs ships small, complete, fairly-priced games on a fast cadence — and helps outside creators finish and launch theirs. No DLC. No microtransactions. Complete games at honest prices. Remember when gaming was fun?
We build by directing AI agents, not by typing code line by line. Your job is to make that produce shippable games — not demos that fall apart in the wild.
How You'll Work (read this part twice)
You will not hand-write most of the code. You'll orchestrate AI agents (Claude Code and similar) to do it. But "doesn't type code" is not "doesn't read code." You are the judgment in the loop: you architect the approach, review what the agents produce, catch them when they're confidently wrong, and refuse to ship what you can't explain. This is the difference between agentic engineering — which works — and vibe coding — which passes a quick test and breaks later. We do the former.
The Role
You alternate between two modes, never both at once:
Build mode: direct AI agents to take a prototype to a finished, shippable game — ~80% in Godot (GDScript), the rest in web tech (JavaScript/TypeScript).
Port mode (after a game ships): get finished titles onto every platform that makes sense.
This is roughly a 50/50 split — you'll spend about as much time porting finished titles as building new ones. The team handles go-to-market, so you stay focused on the game and the build.
What You'll Do
In build mode:
Direct AI agents to build games in Godot (primary) and web/JS (occasional)
Architect the approach and write tight specs/constraints before the agents start
Read, review, and debug everything the agents produce — own it as if you'd typed it
Run adversarial cross-checks (e.g. a second model reviewing significant changes) and fix what they surface
Polish, harden, and ship
In port mode:
Godot titles: desktop (Windows/Mac/Linux) and mobile via Godot export; console (Switch 1 & 2 / PS5 / Xbox Series) via W4 Consoles middleware
Web titles: desktop via Electron/Tauri, mobile via Capacitor, plus Steam and Apple TV
Handle platform integration (controllers, achievements, saves), certification, and store submission
For a web title that earns a console release: scope and coordinate a native rebuild or an external porting partner (no push-button path exists for web ? console)
Must Have
You're a gamer. You play, you have taste, you know what "finished and fun" feels like.
Hands-on agentic AI development experience. You've actually shipped by orchestrating AI coding agents — not just chatted with a model.
You can read, review, and debug Godot / GDScript fluently. You don't have to love typing it, but you must catch a bad implementation, reason about architecture, and explain any line the AI wrote. If you can't read it, you don't ship it — and we treat "the AI wrote it and nobody understands it" as a fireable risk, not a feature.
At least one shipped game. Any size — you've crossed the finish line before.
Comfortable with build pipelines (GitHub Actions / CI-CD, multi-platform builds).
Nice to Have
Web game experience (JavaScript/TypeScript, Electron/Tauri/Capacitor) for the ~20% non-Godot titles
Prior console work (W4 Consoles; certification for Nintendo / Sony / Microsoft)
Mobile and Apple TV deployment
Familiarity with agent orchestration workflows and
You'll Thrive Here If
You actually finish things
You think a complete game at a fair price is a moral stance, not just a business model
"Remember when gaming was fun" makes you nod
APPLY HERE:
