In this project, I did everything AI-wise. I designed and implemented a modular Utility AI decision-making system in C++ for Unreal Engine to control dynamic enemy behavior in the game.

UtilityAI plugin

In this project, I developed a custom Utility AI system that controls how enemies make decisions in real time. Rather than relying on predictable state machines, this system allows enemies to evaluate their surroundings and dynamically choose the best action based on the current situation — for example, deciding whether to attack, to dash, or reposition. The goal was to create smarter and more reactive enemies that felt less scripted and more adaptable to player behavior. The AI continuously weighs multiple factors, such as distance to the player, health status, or whose player dealt the most damage to them, and uses these to calculate which action makes the most sense at any given moment. This approach makes each encounter feel more natural, since enemies can react differently depending on context rather than following a fixed sequence of steps.

The system was built entirely in C++, but designed with Blueprint extendability, allowing designers to prototype or tweak behaviors without touching code. It’s fully data-driven and reusable, so new AIs can be defined simply through configuration rather than hard-coding logic for each one. It also uses an event-based flow — when an action finishes, delegates and callbacks automatically trigger a new decision, keeping the AI loop continuous and autonomous

Enemy Diversity