Senior Researcher - AI Agents

Overview

Microsoft Research (MSR) AI Frontiers is seeking a Senior Researcher to join our team, focusing on ambitious research to advance AI in areas such as modeling, algorithms, agents, and agentic systems. We foster a vibrant, multidisciplinary environment with an open publication policy and strong ties to academia. Our innovations create real-world value by regularly releasing open-source models, libraries, and tools, and by integrating AI technologies across various Microsoft products.

We are looking for exceptional candidates to push the boundaries of agentic systems and ecosystems. Our current relevant projects include agentic systems like Magentic-UI and Magentic-One, ecosystems such as Magentic Marketplace, frameworks like AutoGen, and models including Phi and Orca, alongside tools like AgentInstruct and AutoGen Studio.

As a Senior Researcher - AI Agents, you will engage hands-on with challenging and impactful projects, applying strong technical skills to build practical solutions. You will also collaborate closely with other researchers and engineers across Microsoft to amplify your impact and grow within a supportive and stimulating environment.

Focus Areas

• Agentic Systems: Building state-of-the-art agentic systems, applications, and environments.
• Agent Ecosystems: Developing agent-native ecosystems (e.g., frontier firms, agentic markets) for effective collaboration between agents and people.
• Advanced Agentic Capabilities: Post-training models for advanced capabilities like reinforcement learning, fine-tuning, self-play, and synthetic data generation.
• Human-Agent Interaction: Creating novel interaction techniques and experiences.
• Evaluation & Tooling: Developing benchmarks, methods, and tooling for agentic capabilities.
• Responsible AI: Ensuring agents operate safely, responsibly, and deliver real value.

Responsibilities

• Outcome Driven Innovation: Strategically identifying unmet needs and creating novel, practical solutions by working backward from the problem.
• Collaborative Innovation: Leveraging expertise within and outside the organization to develop novel products, processes, or research streams.
• Problem Solving: Identifying issues, reviewing information, and developing effective solutions.
• Decision Making: Making informed decisions rapidly, considering alternatives, resources, costs, and tradeoffs.
• Scientific Method: Applying empirical methods for knowledge acquisition, including observation, hypothesis formulation, experimentation, and evaluation.

Qualifications

Required:

• Doctorate (or currently pursuing) in Computer Science or a related field, OR equivalent experience.

Preferred:

• Publication record as a lead author or essential contributor at top venues (e.g., CHI, NeurIPS, UIST, ICML, ICLR, ACL, EMNLP, CVPR, AAAI, ICAPS).
• Doctorate in Computer Science or a relevant field AND 2+ years of related research experience, OR equivalent experience.
• Hands-on experience with large foundation models (e.g., OpenAI GPT models, LLAMA) and state-of-the-art AI/ML frameworks and toolkits (e.g., PyTorch, TensorFlow, LangChain, AutoGen).
• Demonstrated software engineering experience in building and deploying prototypes, applications, or open-source technologies (GitHub profile/code samples encouraged).
• Hands-on experience evaluating AI models or systems (e.g., benchmark experiments, user studies, data analysis).
• Experience working in multi-disciplinary teams with strong communication, collaboration, and feedback skills.
• Reference letters are welcomed, though not mandatory.