Game AI Designer, Decision-Making Systems

Onebrief
Battle Road DigitalPosted 21 February 2026

Job Description

About OnebriefOnebrief is collaboration and AI-powered workflow software designed specifically for military staffs. By transforming this work, Onebrief makes the staff as a whole superhuman - meaning faster, smarter, and more efficient.We take ownership, seek excellence, and play to win with the seriousness and camaraderie of an Olympic team. Onebrief operates as an all-remote company, though many of our employees work alongside our customers at military commands around the world.Founded in 2019 by a group of experienced planners, today, Onebrief’s team spans veterans from all forces and global organizations, and technologists from leading-edge software companies. We’ve raised $320m+ from top-tier investors, including Battery Ventures, General Catalyst, Sapphire Ventures, Insight Partners, and Human Capital, and today, Onebrief is valued at $2.15B. With this continued growth, Onebrief is able to make an impact where it matters most.Battle Road Digital, part of Onebrief, applies state-of-the-art innovations and experience from the video game industry to build the future of simulations, modeling, and operations with AtomEngine – our proprietary cloud-based game engine that’s making the world playable.About the RoleWe're hiring a Game AI Designer to help design and productize the entity “brain” that governs behavior across AtomEngine, enabling both internal teams and external developers to author, configure, and extend intelligent behavior. You're turn complex objectives, constraints, and rules into credible and explainable behavior.This role focuses on decision-making systems, not machine learning. You will work with Hierarchical Task Networks (HTNs), Goal-Oriented Action Planning (GOAP), and state machines to define how entities perceive the world, evaluate options, select actions, execute them, and adapt to dynamic conditions.In addition to designing the behavior logic itself, this role is responsible for productizing the AI system - ensuring it can be safely, clearly, and powerfully exposed to external developers and end users via APIs and the Modkit.What You'll DoDevelop a deep understanding of the simulation’s features, constraints, and emergent dynamicsResearch military doctrine and commander intent to understand the expectations for how entities should behave under a broad range of circumstancesConceptualize and design AI behavior models using HTNs, GOAP, state machines, and related decision frameworks that can express a wide range of intended outcomes and behaviorsDefine the structure, abstractions, and constraints of the entity “brain,” in close collaboration with engineeringTranslate high-level intent (doctrine, rules, roles, objectives) into concrete, inspectable, and configurable decision logicDesign and author entity behavior using a visual scripting editorDesign behavior systems that support variation, parameterization, and reuse, rather than hard-coded outcomesParticipate in iteration cycles to refine behaviors based on simulation outcomes, user feedback, and observed emergent behaviorWhat We Look For3 – 5 years’ experience designing game-based AI/decision-making systems using HTNs, GOAP, state machines, or comparable frameworksProven ability to conceptualize AI behavior models that express a wide range of intended outcomes, constraints, and variationsStrong systems-thinking skills, with the ability to reason about complex decision spaces, tradeoffs, and emergent behaviorExperience partnering closely with engineers to design and evolve runtime AI systems, including defining abstractions, constraints, and interfacesSufficient experience writing code so that you can read, understand, and contribute to behavior-related code, prototypes, or tools in collaboration with engineeringAbility to clearly articulate design intent, assumptions, and limitations to both technical and non-technical stakeholdersDue to federal contract requirements, U.S. Citizenship is mandatory for this position.Preferred QualificationsActive or p ... (truncated, view full listing at source)