New opportunity

Senior Data Scientist

Razer · Singapore

About this role

Razer seeks a Senior Data Scientist to develop AI-driven tools and experiences for game developers and gamers. The role emphasizes reinforcement-learning agents, generative AI agents and retrieval-augmented generation (RAG) to improve game testing, content creation, player interactions and engagement.

Responsibilities

Research, design and implement rewards-based reinforcement-learning agents for testing gaming applications. Develop agents and RAG solutions for dynamic content and adaptive player experiences.
Build and optimize AI-powered gaming tools, develop and fine-tune models, and preprocess game data to improve in-game decision-making. Work with AI engineers, game developers, designers and software engineers to integrate models into development workflows.
Evaluate performance and model drift through experiments and benchmarking; improve efficiency and scalability. Assist with production deployment, monitoring and maintenance. Track advances in gaming AI and document methods and findings for internal sharing and, where applicable, external publication.

Requirements

A Master’s degree or PhD in a relevant field such as computer science, AI or machine learning, plus at least 1 year of AI research or applied machine-learning experience.
Strong background in reinforcement learning, generative AI agents and RAG; proficiency in Python and experience with PyTorch or TensorFlow. Familiarity with large language models, transformers, diffusion models and self-learning AI systems; experience training and fine-tuning large-scale models and deploying AI through AWS, GCP or Azure.
Strong problem-solving and written and verbal communication skills, with an understanding of game mechanics, AI-driven storytelling and player engagement.

Preferred

Experience in gaming or interactive media and with Unity or Unreal Engine. The role is based in Razer’s Singapore office and may require occasional travel of up to one trip per year for conferences, research collaborations or business meetings.

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

Skills for this role

Reinforcement LearningGenerative AIAI AgentsRetrieval-Augmented GenerationPythonPyTorchTensorFlowDeep LearningLarge Language ModelsTransformersDiffusion ModelsModel Fine-TuningData PreprocessingModel EvaluationModel MonitoringAWSGoogle Cloud PlatformMicrosoft AzureUnityUnreal EngineProblem SolvingCommunication

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

RL agents for game testing
The role researches, designs and implements rewards-based agents to test gaming applications. Prepare to explain how you would define rewards, evaluate agent behavior and determine whether testing finds useful gameplay issues.
Generative agents and RAG for player experiences
The posting calls for agents and RAG solutions for dynamic content and adaptive player interactions. Prepare to discuss how game mechanics, storytelling and player engagement would inform design and evaluation.
Game-data preparation and model improvement
The role analyzes and preprocesses game data, then develops and fine-tunes models to improve in-game decision-making. Prepare to describe how data quality and model changes would be assessed.
Integration into development workflows
Razer expects collaboration with AI engineers, game developers, designers and software engineers to build usable gaming tools. Prepare to explain how you would translate research into a tool and communicate design choices across those groups.
Experimentation and production follow-through
The posting specifies benchmarking performance and drift, improving efficiency and scalability, assisting with deployment and monitoring, and documenting findings. Prepare to connect experimental results with production decisions and clear documentation.

Likely challenges to think through

Reward design versus useful test coverage
A likely challenge is that an RL agent could maximize its reward without exercising the game behaviors developers need tested; this is a preparation hypothesis, not a reported issue at Razer. Prepare to discuss reward checks, diverse test scenarios and evaluation beyond reward totals.
Adaptation versus coherent game experiences
Dynamic content and player interactions may require balancing variation with consistency in game mechanics and storytelling; the posting does not specify a current problem. Prepare to explain how you would evaluate generated or retrieved outputs with developers and designers.
Research quality versus production constraints
Advanced models may perform well in experiments yet be difficult to scale or maintain; this is inferred from the stated benchmarking and deployment duties. Prepare to discuss trade-offs among performance, efficiency, drift monitoring and maintainability.

Practice your answers

Technical practice

Suggested practice: How would you design a rewards-based RL agent to test a gaming application?

What to coverThe role calls for RL agents that test gaming applications. Discuss the agent’s observations, actions and reward design, including how you would avoid rewarding behavior that misses important gameplay failures.

Suggested practice: How would you build a RAG solution for dynamic game content?

What to coverThe posting names RAG-based solutions for content creation and adaptive experiences. Explain what content the system would retrieve, how retrieval would inform generation, and how you would assess relevance and consistency.

Suggested practice: How would you design a generative agent for adaptive player interactions?

What to coverThe role connects generative agents with player engagement, game mechanics and AI-driven storytelling. Discuss how an agent could respond to game state and player actions while preserving a coherent experience.

Suggested practice: How would you prepare game data for in-game decision-making models?

What to coverThe posting calls for analyzing and preprocessing game data before training models. Describe how you would check data quality, represent gameplay context and prevent evaluation data from leaking into training.

Suggested practice: When would you use an LLM, transformer-based model or diffusion model in a gaming tool?

What to coverThe posting requests familiarity with LLMs, transformers and diffusion models. Compare their suitability for different content or interaction tasks, and explain how you would test whether a more complex model adds value.

Suggested practice: How would you fine-tune a large-scale model for a gaming application?

What to coverThe role includes model development and fine-tuning, with Python and PyTorch or TensorFlow listed as relevant tools. Outline a training and validation approach, then discuss the trade-offs between adaptation quality and compute cost.

Suggested practice: How would you benchmark an AI gaming tool and detect model drift?

What to coverExperimentation, benchmarking and drift evaluation are stated duties. Describe useful baselines and measures for the tool’s intended task, plus how you would distinguish a real performance change from variation in test data.

Suggested practice: How would you improve an AI gaming tool’s efficiency and scalability?

What to coverThe posting explicitly calls for improving both. Explain how you would locate a bottleneck and weigh faster responses or greater throughput against model quality and resource use.

Suggested practice: How would you support deployment and monitoring of a gaming AI model?

What to coverThe role includes production deployment, monitoring and maintenance, and lists experience with AWS, GCP or Azure. Discuss what you would monitor after release and how you would investigate a decline in model performance.

Suggested practice: How would you integrate an AI model into a game development workflow?

What to coverThe role involves working with developers, designers and engineers to integrate models; Unity or Unreal Engine experience is a plus, not a stated requirement. Discuss interface and testing decisions that would let the team use and evaluate the tool without disrupting its workflow.

Behavioural practice

Suggested practice: Tell me about a time you collaborated across technical and creative disciplines.

What to coverThe role involves AI engineers, game developers, designers and software engineers. Give a real situation, the actions you took to align different needs, and the resulting decision or outcome.

Suggested practice: Tell me about a time you turned research into a practical solution.

What to coverThe posting emphasizes translating AI research into usable gaming tools. Describe the problem and constraints, what you chose to implement, and what the result showed.

Suggested practice: Tell me about a time an experiment challenged your initial approach.

What to coverThe role calls for experimentation, benchmarking and problem-solving. Explain the situation, how you investigated the evidence and changed course, and the outcome.

Suggested practice: Tell me about a time you communicated complex model findings to different audiences.

What to coverWritten and verbal communication, documentation and internal knowledge sharing are stated parts of the role. Describe the audience and situation, how you adapted your explanation, and what decision or understanding resulted.

Suggested practice: Tell me about a time you assessed a new AI approach before adopting it.

What to coverThe role includes tracking advances in gaming AI while building practical solutions. Give a real example of the situation, how you tested the approach against the existing option, and what your assessment led to.

Sharpen your knowledge

Topics to prepare

Reward design for game-testing agents
The posting calls for rewards-based reinforcement-learning agents to test gaming applications. Revision exercise: sketch states, actions, rewards and failure cases for an agent testing a simple game, then explain how you would detect reward hacking.
Generative agents and RAG
The role emphasizes agents and RAG for dynamic content and adaptive player interactions. Revision exercise: diagram an agent that retrieves game lore before generating dialogue, and explain when retrieval helps and how you would assess consistency.
Model families and adaptation
The posting asks for familiarity with LLMs, transformers, diffusion models and self-learning AI systems, plus experience training and fine-tuning large-scale models. Revision exercise: compare which of these approaches could support dialogue versus visual content, and outline a fine-tuning decision.
Game mechanics and player experience
The role calls for understanding game mechanics, AI-driven storytelling and player engagement. Revision exercise: take one hypothetical game mechanic and explain how an adaptive agent might change the player experience, including an undesirable outcome to guard against.
Evaluation, drift and scalability
The posting explicitly calls for experiments, benchmarking, drift evaluation, efficiency and scalability. Revision exercise: design a comparison between two versions of a game AI model, specifying quality measures, a drift signal and a resource-cost measure.

Skills to revise

Implement an RL test agent
The posting requires Python and experience with PyTorch or TensorFlow for deep learning. Practice suggestion: use Python and one of those frameworks to build a small rewards-based agent in a simplified game-testing scenario; inspect unsuccessful runs and revise the reward or training setup.
Prototype a retrieval-backed interaction
The role calls for advanced RAG-based solutions for game content and interactions. Practice suggestion: create a small set of fictional game-lore passages, retrieve relevant passages for player prompts, and check whether generated responses stay consistent with the retrieved material.
Prepare game data and fine-tune a model
The posting includes preprocessing game data and developing and fine-tuning models for in-game decision-making. Practice suggestion: define a small hypothetical game-data sample, document cleaning and train/test separation, then describe or run a limited fine-tuning experiment and compare it with a baseline.
Benchmark and monitor a model
The role involves evaluating performance and drift and assisting with production monitoring. Practice suggestion: write a repeatable evaluation script for a hypothetical gaming model, record results across two data slices, and specify what change would trigger investigation.
Explain a deployment handoff
The posting asks for cloud-based AI deployment experience with AWS, GCP or Azure and collaboration with engineers and game developers. Practice suggestion: outline how you would hand off a model for deployment on one of those platforms, covering inputs, outputs, monitoring and maintenance responsibilities; this outline is preparation, not a stated Razer process.

Your preparation checklist

Prepare an RL testing walkthrough
The role calls for rewards-based agents that test gaming applications. Prepare a project walkthrough, or a clearly labeled design example, covering the game environment, reward design, training approach, failure cases and testing results.
Sketch a generative-agent and RAG design
The posting names dynamic content and adaptive player interactions. Draw an illustrative architecture showing what an agent generates, what RAG retrieves, how game context enters the system and how outputs would be evaluated.
Gather model-building evidence
Prepare concrete examples of Python work with PyTorch or TensorFlow, game-data preprocessing, and model training or fine-tuning. Be ready to distinguish hands-on experience from familiarity with LLMs, transformers, diffusion models and game engines.
Build an evaluation and production plan
The role includes benchmarking, drift evaluation, scalability, cloud deployment and monitoring. Draft a short plan with baselines, experiment metrics, drift signals and production checks; label proposed choices as your suggestions.
Rehearse a cross-functional explanation
Practice explaining one AI tool to a game developer or designer and an AI engineer. Show how it could fit a development workflow and how game mechanics, storytelling or player engagement would shape trade-offs.
Write focused questions for Razer
Ask which gaming applications are priorities for RL testing, how the team evaluates generated content and player interactions, and how AI engineers, developers and designers divide integration and production responsibilities.

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…