The AI Platform organization at Microsoft builds the end-to-end Azure AI stack/PaaS and is core to Azure’s innovation and differentiation, as well as all of Microsoft’s flagship products, from Office to Teams, to Xbox. We are the team building Azure OpenAI, Azure ML, Cognitive Services, and the global Azure AI infrastructure for running the largest AI workloads on the planet. Within AI Platform, the AI Foundry team enables data scientists and developers to quickly and easily build, train, deploy, manage, and consume machine learning model. Our AI Foundry OSS Model Customization Team is at the forefront of this mission working on groundbreaking projects for customizing of OSS models. We collaborate closely with research institutions, industry leaders, and organizations worldwide to create innovative solutions that impact millions of users. As a Senior Software Engineer in our team, you will: Write clean and concise code with unit tests. Collaborate with researchers and data scientists to implement model customization techniques, including Finetuning, Reinforcement Finetuning, Distillation. Optimize model performance, scalability, and efficiency. Conduct experiments to evaluate model performance, robustness, and generalization. Explore novel techniques and approaches to enhance model capabilities.
Production code in internet scale services and distributed systems (experience in Software Engineering, 5 years)
Cloud platforms (Azure, AWS) and distributed computing (Kubernetes) (experience in Cloud Computing)
Transformer-based and diffuser-based models (GPT, Llama, Stable diffusion) (experience in Machine Learning)
Preferred Qualifications
Transformer-based and diffuser-based models (GPT, Llama, Stable diffusion) (experience in Machine Learning)
Responsibilities
Write clean and concise code with unit tests
Collaborate with researchers and data scientists to implement model customization techniques, including Finetuning, Reinforcement Finetuning, Distillation
Optimize model performance, scalability, and efficiency
Conduct experiments to evaluate model performance, robustness, and generalization
Explore novel techniques and approaches to enhance model capabilities