InfoQ launches AI safety certification cohorts focusing on data protection and coding agent oversight

InfoQ is opening its doors to two separate five-week online cohorts this October for practitioners seeking deeper expertise in AI system safety. AI Security & Privacy Engineering begins October 26, while AI-Assisted Engineering kicks off October 19. Each blends hands-on learning with peer feedback from experienced engineers across different industries.

Protecting sensitive data in AI workflows#

The security-focused cohort teaches participants to trace the flow of sensitive information through an AI system from entry point to output. Using threat modeling and red team exercises, attendees map architectural vulnerabilities and design testing approaches for security controls. During the final project, cohort members evaluate a real AI product architecture, document identified risks, justify the controls they'd implement, outline testing strategies, and clarify accountability for remaining risk.

Katharine Jarmul, author of Practical Data Privacy, facilitates this track. Her expertise spans privacy and security practices in machine learning environments. "An AI security review has to follow the data and the decisions across the whole system," she explains. "In the cohort, we'll map where sensitive information can go, test the controls we choose, and make clear who owns the risks that remain."

Building guardrails for autonomous code changes#

The second cohort addresses a different challenge: ensuring that AI coding agents make safe, auditable modifications to existing codebases. Participants work with a shared legacy repository, learning to construct proper context boundaries for the agent, restrict its capabilities, introduce verification tests, and establish human review gates before automated checks shift to CI/CD pipelines. Their capstone involves presenting the complete verification framework they've built and proposing new agent rules based on five weeks of logged behavior.

Zichuan Xiong, Head of AIOps at Thoughtworks, co-leads this program alongside Premanand Chandrasekaran, Head of Technology at Thoughtworks. Xiong frames the core tension: "A coding agent can make a change quickly, but the harder question is what it was allowed to do and how we know the change is sound. We'll build context and limit permissions before putting checks around an agent working in an existing codebase."