That Ubisoft conversation gave me something I didn’t expect: clarity on what “game AI developer” actually means at a studio level. These aren’t NPC scripting gigs. Studios like EA, Riot, PlayStation, Ubisoft are hiring for roles like AI Engineer and Tools Engineer, Bots & Simulations. People building testing bots, QA bots, in-game agents that play with the player or against them. Specific work, and the job titles are nothing you’d think to search for.
So I pulled every game AI listing I could find and read each one like a requirements doc.
The skills overlap with what you’re already doing more than you’d think. Python, decision systems, state machines, goal-based agents, autonomous systems that make choices under pressure. If you’ve been building a bot that plays on the ladder, you’ve been practicing most of this without knowing the industry had a name for it.
One skill kept showing up that I didn’t have. Reinforcement learning. EA SEED, PlayStation, Riot, it was in every one of them. PiG_Bot has zero of it. That was my gap.
My first instinct was to feel behind, like maybe I needed to scrap what I had and start fresh with an RL textbook. Except, don’t do that. The listing told me exactly what to learn next, and my bot is the project I learn it on.
If your gap is RL, sentdex did a reinforcement learning video for StarCraft 2 which is a solid starting point. It’s dated but the core concepts hold. Whatever resource you find, your bot is the testbed. You don’t need a new project, you need a new chapter on the one you already have.
