Senior Researcher - Foundations of Generative AI- Microsoft Research

Microsoft ResearchAI Frontierslabis seeking applications for the position ofSeniorResearcher - Foundations ofGenerativeAIto join their team inNew York, NY.

Themission of theAI Frontierslabis to expand the pareto frontier ofArtificial Intelligence(AI)capabilities, efficiency, and safety through innovations in foundation modelsand learning agent platforms.Some of our projectsinclude work onsmalllanguage/actionmodels(e.g.,Phi, Orca, Fara-7B),newarchitecturesandoptimizers(e.g., Belief State Transformer,Dion),and agentic AI systems(e.g.AutoGen,MagenticOne,OmniParser).

We areseekingaSeniorResearcher- Foundations ofGenerative AIto join our teamandlead efforts indiscoveringand building the foundations ofgenerativeAI through representations andobjectives.As aSenior Researcher - Foundations of Generative AI, you willplay a crucial role in leading, developing,improving,andexploring new architectures,representations, andlearningobjectivesthatunlocknew capabilities and/or scalability.Your work will have a significant impact on thedevelopmentofcutting-edgetechnologies, advancingthestate-of-the-art,andproviding practical solutions to real-world problems.

Our ongoing research areasencompassbut are not limited to:

• Newarchitectures andlearning methodologiesto enablealways-on,proactiveagents
• Test time training to support infinite context windows
• Active visual reasoning for orders-of-magnitude faster image processing/encoding
• World models of user/system interactions to enable imagination rollouts and search
• Multi-scale temporal reasoning and planning in transformer models
• Continual learning and adaptation methods that operate at human speed
• New multimodal model architectures and training methods for Vision-Language (VLM) and Vision-Language-Action (VLA) models
• Proactive, real-time agents for computer use, racing, and gaming

TheMicrosoft ResearchAI Frontiers laboffers a vibrant environment forcutting-edge, multidisciplinary research, including access to diverse, real-world problems and data, opportunities for experimentation and real-world impact, an open publication policy, and close links to top academic institutions around the world.

Microsoft's mission is to empower every person and every organization on the planet to achieve more.Asemployeeswe come together with a growth mindset, innovate to empower others, and collaborate to realize our shared goals. Each day we build on our values of respect, integrity, and accountability to create a culture of inclusion where everyone can thrive at work and beyond.

#MSR

• Applyresearch andengineering skillstodevelop,prototype, andevaluatecutting-edgeresearch ideas.
• Work closely with other researchers and engineers to rapidly prototype and test new research ideas, driving a high-impact agenda and publishing results where appropriate.
• Collaborate hands-on with other researchers, engineers, and internal and external product groups to deliver high-impact solutions to real-world problems.
• Embody our cultureandvalues.

• Doctorate (or currently pursuing)in Computer Scienceor relevantfield
  • ORequivalent experience.

Preferred Qualifications

• Doctoratein Computer Scienceor relevantfieldAND 2+ years relatedresearchexperience
  • ORequivalent experience.
• Research program demonstrated bypublic artifacts like models, tools, code in the AI space orpublications at the following conferences: NeurIPS, ICML,ICLR, ACL, NAACL, CVPR, COLT,ECCV,ICCV,EMNLP.
• 2+ years of academic or industryexperience in developing, applying, and/or implementing algorithms for machine learning/statistics, using common ML engineering programming languages and platforms such as Python, Python numerical libraries,PyTorch, TensorFlow and/orHuggingFace.
• Experience publishing academic papers as a lead author or essential contributor in a top AI conference or journal.
• Deep understanding of frontier model architectures, especially transformers and state space models
• Hands-on experience building and working with Large Language Models (LLMs) or multimodal models (VLMs, VLAs), including pre-training, fine-tuning, and inference
• 2+ years of industryor academicexperience with building,debuggingandoptimizinglarge-scale ML training pipelines.
• Demonstrated software engineering excellence building and deploying prototypes, applications, or open-source (OSS) technologies. Providing a link to a GitHub profile and/or code samples on your CV/resume, is highly encouraged.
• Ability to work independently and ramp-up quickly on complex projects or unfamiliar code
• Ability to collaborate, communicate effectively, and work as part of a multi-disciplinary team
• Keen interest in real-world applications and impact.

#MSR