Inoutbox
Overview
We are developing a psychologically driven horror game with systemic, emergent enemy behavior. Important note that this is multiplayer game
Enemies are not scripted encounters. They are layered AI entities built from independent behavior modules, explicit sensor systems, and environmental rules such as light, observation, footsteps, and player states.
We are looking for a Senior AI Engineer with deep specialization in enemy behavior architecture, not a general gameplay programmer. Additional BIG plus will be multiplayer experience
Role & Responsibilities
You will design and implement a scalable enemy AI architecture that supports:
3 Enemy types
2 variants or modifier of each enemy type
4 Global level modifiers
Explicit sensor channels: Vision, Sound, Steps
A fully separate, query-based Light System
Key Responsibilities
Design modular AI architecture with independent layers
Translate psychological and abstract behavior concepts into gameplay logic
Implement complex state-driven and rule-driven enemy behaviors
Ensure deterministic behavior during cutscenes and scripted moments
Build anti-exploit systems using delays, hysteresis, and cooldowns
Collaborate closely with game design on behavior tuning and edge cases
Behavior Systems You Will Work On
Observation-based behaviors (freeze, slow down, sudden charge)
Light-driven logic (attraction, avoidance, anchoring to light zones)
Footstep-only detection separate from noise or voice
Possession mechanics where players become threats to teammates
Non-standard traversal (wall-walking, ceiling movement, alternative navigation)
Core Technical Requirements (Must-Have)
You must have strong experience with:
Enemy AI architecture supporting complex behavior combinations
Finite State Machines and/or Behavior Trees with custom extensions
AI perception systems with explicit sensor separation
LOS and FOV systems with dynamic modifiers
Debuggable and inspectable AI systems with clear states and transitions
Important Notes
This role focuses on AI systems, not scripting individual encounters
We do not use monolithic behavior trees or hard-coded per-level logic
Light, perception, and behavior layers must remain independent
Timezone note:
We are currently only considering candidates based in North America, South America, or Europe due to timezone overlap
1v1Me
1v1Me
1v1Me
Aftershock Media Group
Voldex Games
Respawn Entertainment