Research Associate - School of Computer Science - S3D

Carnegie Mellon University•Pittsburgh, PA

2d•Onsite

About The Position

The Software and Societal Systems Department (S3D), in the School of Computer Science (SCS) at Carnegie Mellon University has two part-time research associate openings to work on an interdisciplinary project focused on modeling and mitigating risks introduced by agentic AI in the enterprise. This opening can be a particularly good match for a motivated person with training in Responsible AI, AI Governance, Cybersecurity and Privacy, including experience with Agentic AI technologies. Agentic AI systems—capable of autonomous task execution, decision-making, and multi-step planning—are rapidly being integrated into workplace environments. These systems can schedule meetings, interact with email and web services, generate reports, access internal systems, and coordinate across third-party tools, effectively acting as semi-autonomous digital employees. While these capabilities offer significant productivity gains, they also introduce fundamentally new categories of risk that extend beyond traditional model-level concerns. In particular, agentic AI expands the risk surface from isolated model behavior to complex, system-level interactions involving tools, data, users, and external services. This project focuses on two central research questions: How does agentic AI—particularly autonomy, tool use, and multi-step planning—reshape traditional cybersecurity and privacy threat models in enterprise environments? What technical and governance mechanisms (e.g., permissioning, monitoring, sandboxing, policy enforcement) are most effective in mitigating these risks while ensuring meaningful human oversight and control?

Requirements

Bachelor’s degree in computer science required
1-3 years of research experience required
Strong background in Privacy and Security
Strong organization and communication skills necessary
Experience with Responsible AI, including Agentic Technology
Experience collaborating with others in the context of research projects
Successful completion of a pre-employment background check

Nice To Haves

Master’s degree in computer science preferred
Experience in HCI and/or threat modeling would be a plus
A combination of education and proven experience from which comparable knowledge is demonstrated may be considered

Responsibilities

Combine the development of threat and failure taxonomies for agentic AI in the enterprise
Development and simulation of synthetic workflows
Failure mode analysis
Development and evaluation of mitigation strategies
Identification/development of relevant metrics