New 
Applied Scientist, Gaming AI
 Microsoft | |
 United States, Washington, Redmond | |
 Aug 10, 2025 | |
|
OverviewWould you like to be at the forefront of AI and Gaming and work with state-of-the-art AI research and the top gaming studios in the world? At Gaming AI, we are exploring emerging technology trends to craft the next era of gaming. We are venturing beyond the horizon and charting a course forward with players and creators at the center. Our goal is to define the future of Xbox by advancing our mission of bringing joy and community to every player on the planet. If you value dynamic and agile teams that are proactively advocating for a diverse workforce, we have a role for you. We are looking for an Applied Scientist, Gaming AI with high quality machine learning and engineering skills, bringing proven problem-solving skills and a passion to apply advanced AI techniques in gaming domains. As a key contributor, you will develop state-of-the-art machine learning techniques to elevate user and content experiences to the next level in gaming. You will collaborate with a diverse global team of engineers, product managers and scientists that develops and applies sophisticated machine learning techniques (e.g., reinforcement learning, computer vision, natural language processing, recommender systems and more) in gaming related products. You'll be expected to bring technical rigor, creativity, and execution to help move research into production.
ResponsibilitiesIn this role, you will collaborate with experts in machine learning and distributed systems to develop advanced AI models that power innovative gaming experiences. You will work on reinforcement learning, computer vision, small and large language models (SLMs and LLMs), and other advanced machine learning multimodal techniques that integrate vision, language, actions, audio, and more. Your ML models will help advance the state of the art in both production and distribution of the gaming industry. Your key responsibilities will include:Designing and developing intelligent, versatile AI models tailored to diverse gaming scenarios, including player understanding, dynamic content generation, and personalized experiences.Applying advanced machine learning techniques to enhance gaming experiences, including:Developing and deploying transformer-based architectures (e.g., GPT, ViT) for both SLMs and LLMs.Implementing reinforcement learning algorithms (e.g., PPO, DQN, A3C).Leveraging computer vision frameworks such as OpenCV and Detectron2.Utilizing a variety of NLP libraries and toolkits.Fine-tuning foundation models for domain-specific tasks, including prompt engineering, data curation, and evaluation.Designing and executing benchmarking strategies and rigorous evaluations to assess model performance across key dimensions-such as accuracy, latency, and robustness-ensuring reliability, effectiveness, and alignment with Gaming AI goals in both offline and live environments.Building and maintaining scalable ML pipelines for training, evaluation, and inference using tools like Python, PyTorch, ONNX, and Azure ML.[VF1] Applying model compression techniques (e.g., quantization, pruning, knowledge distillation) to optimize inference performance.Conducting A/B testing and causal inference analysis to evaluate the impact of AI-driven features on player engagement and retention.Driving data strategy, including acquisition, generation, and preparation to support model development and evaluation.Contributing to production readiness, including model training, fine-tuning, quantization, and integration into end-to-end systems and pipelines.Document approaches, share learnings, and contribute to a culture of responsible and grounded innovation in AI.Collaborating across Xbox and partner teams to support the adoption and advancement of AI methodologies, and to help deliver key AI-powered features across the Xbox ecosystem.Working closely with product and engineering partners to scope problems, define success metrics, and deliver robust models that integrate into game or platform workflows.Thinking beyond the model: designing algorithms and systems that embed ML models into real-time, scalable, and user-facing applications-balancing scientific rigor with engineering pragmatism. | |