New opportunity

Research Fellow (HCI, Generative AI)

National University of Singapore · Singapore

About this role

The NUS Immersive Reality Lab seeks a Research Fellow to design, develop and evaluate AI-assisted exhibition tools and immersive heritage applications. Based at Kent Ridge Campus in the College of Design and Engineering’s Electrical and Computer Engineering unit, the fellow will collaborate with curators, designers and heritage professionals on systems for exhibition planning and public engagement.

Responsibilities

Design user-centered interactions and voice- and text-prompting interfaces for controlling spatial layouts, narrative flow and iterative design. Build prototypes, conduct user studies and evaluate usability.
Develop prompt-driven 3D scene and gallery generation pipelines, integrate diffusion models for visualization and content creation, support text-to-scene and voice-to-scene workflows, and implement systems in real-time engines such as Unity or Unreal.
Build AI-assisted exhibition prototyping and visualization platforms. Develop and document end-to-end workflows covering prompt engineering, model fine-tuning and iteration, asset management and versioning, and spatial generation and annotation.

Qualifications

PhD in HCI, computer science, engineering, AI or a related field.
Strong expertise in at least one of: HCI, including user studies, UX or interaction design; generative AI, including diffusion models, and 3D reconstruction and rendering techniques such as Gaussian Splatting and NeRF; or immersive AR, VR or MR systems.
Demonstrated research capability through publications or substantial project or portfolio experience, for example in areas represented by CHI, SIGGRAPH, NeurIPS, TVCG or IEEE VR.

The posting states that the role is open to a fixed-term contract.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

Human-Computer InteractionGenerative AIMachine LearningUX ResearchInteraction DesignUser StudiesUsability EvaluationMultimodal PromptingDiffusion Models3D Reconstruction3D RenderingGaussian SplattingNeRFUnityUnreal EnginePrompt EngineeringModel Fine-TuningAugmented RealityVirtual RealityMixed RealityPrototypingAsset Management

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

User-centered exhibition controls
Design voice- and text-prompting interfaces that let users shape spatial layouts, narrative flow and iterative designs; prototype them and evaluate usability through user studies.
Generative 3D and immersive implementation
Develop prompt-driven scene and gallery generation, integrate diffusion models for visualization and content, and support text-to-scene and voice-to-scene workflows in real-time engines such as Unity or Unreal.
End-to-end exhibition platform
Build AI-assisted prototyping and visualization tools, with documented workflows for prompting, model iteration, asset management and versioning, and spatial annotation.
Research grounded in heritage practice
Collaborate with curators, designers and heritage professionals on exhibition planning and public engagement. The role also calls for demonstrated research capability through publications or substantial project or portfolio work.

Likely challenges to think through

Creative control versus generated output
A likely design challenge is making generated scenes easy to revise without losing spatial or narrative intent. Prepare to explain how you would test whether voice and text controls give users predictable, useful results.
Real-time performance versus visual quality
Generative visualization and immersive implementation may require trade-offs between scene fidelity, responsiveness and iteration speed. Prepare an example of how you would choose what to generate, optimize or defer in a prototype.
Research rigor versus partner usability
Working with heritage professionals may require balancing measurable study outcomes with tools that fit exhibition-planning practice. Prepare to describe how you would turn partner feedback into design changes and evaluate whether those changes help.

Practice your answers

Technical practice

How would you design an interface for curators to control gallery layouts and narrative flow?

What to coverThe role calls for user-centered controls for spatial layouts, narrative flow and iterative design. Suggested preparation: explain how you would discover users’ needs, make generated changes inspectable, and let users revise them.

How would you evaluate whether an AI-assisted exhibition prototype is usable?

What to coverThe posting calls for prototyping, user studies and usability evaluation. Suggested preparation: describe suitable study tasks, what you would observe or measure, and how findings would guide the next iteration.

How would you support both voice-to-scene and text-to-scene interaction?

What to coverThe role includes voice and text prompting for scene creation. Suggested preparation: discuss how users would review or correct an interpreted request, and the trade-offs between spoken and typed input.

How would you structure a prompt-driven 3D gallery-generation pipeline?

What to coverThe posting calls for prompt-driven scene and gallery generation. Suggested preparation: trace a request from prompt interpretation through spatial generation to a reviewable scene, identifying where constraints and human edits belong.

Where would diffusion models fit in an exhibition-visualisation workflow?

What to coverThe role includes diffusion models for visualisation and content generation. Suggested preparation: distinguish generating visual content from arranging a coherent 3D space, and explain how you would assess whether outputs suit the exhibition brief.

When might Gaussian Splatting or NeRF be useful for an immersive heritage application?

What to coverThe qualifications name these as examples of 3D reconstruction and rendering techniques, rather than requiring both. Suggested preparation: compare their potential uses and practical trade-offs for representing a heritage scene, including rendering and editability considerations.

How would you bring a generated gallery into a real-time engine?

What to coverThe posting names Unity and Unreal as examples for implementation. Suggested preparation: explain how you would handle scene assets, interaction and performance while preserving the ability to revise a gallery design.

How would you decide between prompt iteration and model fine-tuning?

What to coverThe documented workflow includes prompt engineering, model fine-tuning and iteration. Suggested preparation: compare the effort and expected benefits of each approach, and describe how you would evaluate changes against the intended visual or spatial result.

How would you manage assets and versions across exhibition prototypes?

What to coverAsset management and versioning are explicit parts of the end-to-end workflow. Suggested preparation: explain how prompts, generated assets, scene changes and annotations could remain traceable so collaborators can review or revisit a design.

How would you design an AI-assisted exhibition prototyping platform around spatial annotation?

What to coverThe role calls for a prototyping and visualisation platform with spatial generation and annotation. Suggested preparation: describe how collaborators could inspect, annotate and revise generated spaces, and what a prototype would need to demonstrate before a user study.

Behavioural practice

Tell me about collaborating with specialists outside your discipline.

What to coverThe role involves working with curators, designers and heritage professionals. Suggested preparation: use a real situation/action/result example showing how you understood another specialist’s goal and translated it into a design or research decision.

Describe a time user feedback changed your prototype.

What to coverPrototyping, user studies and iterative design are stated responsibilities. Suggested preparation: give a real situation/action/result example that separates what participants did or said from the changes you made and the outcome you observed.

Tell me about resolving competing priorities in an interactive system.

What to coverThe role combines usable interactions, AI generation and immersive implementation. Suggested preparation: use a real situation/action/result example to explain the trade-off, whose input you sought, and how you assessed the decision.

Describe a time you made an experimental workflow understandable to collaborators.

What to coverThe posting calls for documenting end-to-end workflows and collaborating across disciplines. Suggested preparation: give a real situation/action/result example of what you documented, how others used it, and what you improved.

Tell me about communicating the contribution of a research project.

What to coverThe posting seeks demonstrated research capability through publications or substantial project or portfolio experience. Suggested preparation: use a real situation/action/result example to explain the research question, your contribution, the evidence gathered and the limits of the findings.

Sharpen your knowledge

Topics to prepare

User-centered HCI evaluation
Review how to turn an exhibition-planning need into a research question, user-study tasks and usability measures. Exercise: outline a study comparing two ways to adjust a gallery layout, including what you would observe and how you would interpret the results. The role explicitly includes user studies and usability evaluation.
Multimodal interaction design
Review how voice and text prompts can control spatial layouts, narrative flow and design iteration. Exercise: map one curator request to an interface response, then identify an ambiguity the user would need to resolve. These interactions are explicit design responsibilities.
Generative 3D methods
Review the roles of diffusion models in visualization and content generation and of Gaussian Splatting and NeRF in 3D reconstruction and rendering. Exercise: explain where each could fit in a prompt-driven gallery workflow and what would need evaluation. These methods appear in the responsibilities or qualifications; the posting asks for strong expertise in at least one listed area, not all of them.
Immersive heritage experience
Review interaction and evaluation considerations for AR, VR or MR in a heritage exhibition. Exercise: sketch how a visitor would encounter and navigate one generated scene. Immersive heritage applications are central to the role, and immersive systems are one of the listed expertise areas.
Research contribution and evidence
Exercise: draft a brief explanation of a research question, method, result and contribution using a publication or substantial project as a template, without assuming any particular candidate experience. The posting explicitly seeks demonstrated research capability through publications or strong project or portfolio experience.

Skills to revise

Prototype a prompting interaction
Exercise: make a small interface prototype in which text or voice input changes a gallery layout or narrative sequence, including a way to revise the result. This practices the requested multimodal interfaces, interaction design and prototyping.
Plan and assess a user study
Exercise: write tasks for a curator or designer to complete with the prototype, a short observation plan and criteria for identifying usability problems. The role explicitly calls for user studies and usability evaluation with exhibition stakeholders.
Trace a scene-generation pipeline
Exercise: diagram the inputs, generated 3D scene, diffusion-based visualization and revision steps for a text-to-scene or voice-to-scene request. This directly rehearses the specified generation pipelines without assuming a particular model or tool.
Implement in a real-time engine
Exercise: outline or build a minimal generated-gallery scene in Unity or Unreal, showing how a layout change would appear during iteration. The posting names these as example engines for implementation; it does not require both.
Document an end-to-end workflow
Exercise: draft a reproducible workflow covering prompt revisions, model fine-tuning and iteration, asset versioning, and spatial annotation. These are explicitly named components of the exhibition prototyping and visualization platform work.

Your preparation checklist

Map evidence to the qualifications
Make a one-page evidence sheet for the stated PhD requirement, research publications or substantial projects, and your strongest area among HCI, generative AI and 3D techniques, or immersive systems. Mark any gaps rather than implying experience.
Prepare a research project walkthrough
Select a relevant publication or project, if available, and prepare a brief account of the research question, your contribution, prototype, evaluation and findings. Connect it to the lab’s stated exhibition or immersive-heritage work.
Sketch a multimodal interaction
As interview practice, draw a voice-and-text interface for changing a gallery’s spatial layout or narrative flow. Show how a curator could review and revise each change; this is a proposed design, not a specified assignment.
Diagram a scene-generation pipeline
Prepare a technical diagram from text or voice input to a 3D scene and real-time preview. Explain where diffusion-based visualisation, spatial generation and a Unity or Unreal implementation could fit, and identify what you would need to validate.
Outline a reproducible platform workflow
Draft a sample workflow covering prompt iteration, possible model fine-tuning, asset management and versioning, and spatial annotation. These stages are named in the posting; the workflow you present should be clearly framed as your approach.
Draft a usability study plan
Write a short plan for testing an exhibition-planning prototype with intended users: tasks, observations, usability measures and how findings would inform iteration. User studies and usability evaluation are explicit responsibilities.
Bring questions for collaborators
Prepare questions about curators’ and heritage professionals’ planning needs, available exhibition assets, evaluation priorities, and how the fellow would work with designers on public-engagement applications.

Availability last checked: 4 Nov 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in Singapore. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…