AI Engineer – Generative Agent-Based Modelling (GABM)

About the Role

We are looking for an AI Engineer for a two-year contract position to develop a Generative Agent-Based Modelling (GABM) platform for large-scale simulation of human behaviour and interactions in complex environments at ST Engineering, AI/Robotics Strategic Technology Centre (AI.R STC). The platform combines large language models, vision-language models, generative world models, AI-generated 3D content, multi-agent systems and 3D simulation to support applications such as public safety, crowd management, emergency response and urban planning.

You will work closely with AI researchers, simulation engineers and domain experts to translate research ideas into scalable, reliable prototypes and operational demonstrations.

Key Responsibilities

• Design and implement autonomous agents powered by large language models, including perception, memory, reasoning, planning, decision-making and action execution.
• Develop multi-agent interaction and coordination mechanisms for realistic individual and group behaviours.
• Build and optimise large-scale simulations involving thousands of concurrent agents.
• Integrate generative agents with 3D simulation environments such as Unreal Engine 5, including navigation, animation, environment perception and event handling.
• Develop AIGC pipelines for generating and adapting simulation-ready 3D characters, objects, scenes and environments from text, image or other multimodal inputs.
• Explore and integrate generative world models for environment creation, scenario generation, state prediction and interactive simulation.
• Apply vision-language models to visual perception, scene understanding, visual grounding, navigation and context-aware interaction between agents and their environment.
• Develop interfaces and data pipelines connecting AI services, simulation engines and external systems using Python, C++ and APIs or messaging frameworks.
• Create agent profiles and behaviour models from scenario requirements, real-world data and domain knowledge.
• Evaluate agent realism, behavioural diversity, scalability, latency and simulation performance using quantitative and qualitative methods.
• Conduct experiments, benchmark alternative models and algorithms, and translate research findings into working systems.
• Collaborate with internal teams, universities and external partners on system integration, demonstrations, technical documentation and project deliverables.
• Contribute to software architecture, code quality, testing, deployment and technical roadmap development.

Requirements

• Master's or PhD degree in Computer Science, Artificial Intelligence, Robotics, Game Technology or a related field.
• Strong programming skills in Python; experience with C++ is an advantage.
• Hands-on experience in at least one of the following areas:
  • Large language model applications or AI agents;
  • Vision-language models, multimodal AI or embodied navigation;
  • Generative AI for 3D content or world modelling;
  • Multi-agent systems or agent-based modelling;
  • Game engines or 3D simulation;
  • Reinforcement learning, planning or behavioural modelling.
• Experience developing production-quality or research prototype software in a collaborative environment.
• Good understanding of software engineering practices, including Git, modular design, debugging, testing and documentation.
• Strong problem-solving skills and the ability to work independently on open-ended technical challenges.
• Good written and verbal communication skills.

Preferred Qualifications

• Experience with Unreal Engine 5, including C++, Blueprints, navigation, animation systems and performance profiling.
• Experience with LLM agent frameworks, retrieval-augmented generation, memory systems, prompt design or local model deployment.
• Experience with vision-language models for scene understanding, visual grounding, navigation, embodied question answering or human-agent interaction.
• Experience with generative world models or AIGC-based 3D generation, including text/image-to-3D, neural rendering, NeRF, Gaussian Splatting, diffusion models or related techniques.
• Familiarity with 3D asset pipelines and preparing generated characters, objects or environments for real-time simulation engines.
• Knowledge of crowd simulation, social-force models, pedestrian dynamics, emergency evacuation or human behaviour modelling.
• Experience optimising simulations for hundreds or thousands of agents through multithreading, distributed computing, GPU acceleration or level-of-detail techniques.
• Familiarity with ROS/ROS2, NVIDIA Omniverse or Isaac Sim, Unity, UnrealCV, ZeroMQ or similar integration technologies.
• Experience with machine learning frameworks such as PyTorch and with deploying models as scalable services.
• Publications, open-source contributions or project experience in generative agents, multi-agent simulation, embodied AI or related fields.

What You Will Work On

• A scalable GABM platform capable of simulating at least 2,000 agents.
• High-fidelity generative agents for key roles, with individual goals, memory, social relationships and adaptive decision-making.
• Simulation scenarios for public safety, crowd movement, evacuation and other complex real-world operations.
• Integration of AI-driven behaviour with realistic 3D environments, character movement and animation.
• AIGC pipelines for rapidly generating simulation-ready 3D assets and scenario environments.
• Generative world-model capabilities for creating, predicting and adapting interactive simulation environments.
• Multimodal agents that use vision-language models to understand scenes, navigate and interact with other agents and the environment.
• Evaluation tools and dashboards for analysing agent behaviour and scenario outcomes.

What We Offer

• The opportunity to work at the intersection of generative AI, multi-agent systems and high-fidelity simulation.
• Collaboration with leading academic and industry partners.
• Direct involvement in applied R&D projects with meaningful real-world use cases.
• A multidisciplinary environment with room to shape the technical direction of an emerging platform.