Research Intern — Red Team, Agentic AI
Research how autonomous AI agents can be manipulated, compromised, and defended as they gain access to tools, data, and real-world systems.
About the role
We are looking for an unusually curious researcher or engineer who wants to explore the security of agentic AI systems. You will work directly with our research team to investigate how AI agents behave when given tools, memory, external data, and the ability to take actions. You will build experimental agents, design adversarial environments, develop proof-of-concept attacks, and measure how different models and architectures respond. This is an exploratory research role: you do not need to know all the answers, but you should be comfortable learning quickly, writing code, breaking systems, and explaining what you discovered. Strong candidates may come from computer science, security, machine learning, mathematics, or another technical discipline. Formal security experience is not required if you can demonstrate exceptional technical ability.
Responsibilities
- Research attack techniques against autonomous and tool-using AI agents.
- Build experimental agent environments and adversarial test cases.
- Investigate prompt injection, indirect prompt injection, tool abuse, memory attacks, data exfiltration, and privilege escalation.
- Develop reproducible proof-of-concept attacks and evaluation harnesses.
- Run experiments across different models, prompts, tools, and system architectures.
- Analyze experimental results and identify meaningful patterns or failure modes.
- Read and reproduce relevant research from academic and industry sources.
- Document findings with clear methodology, evidence, and conclusions.
- Work closely with senior researchers to turn promising observations into deeper investigations.
Requirements
- Currently pursuing or recently completed a degree in computer science, machine learning, cybersecurity, mathematics, or a related technical field.
- Strong programming ability, preferably in Python.
- Genuine interest in AI systems and security research.
- Ability to learn unfamiliar technologies quickly.
- Strong analytical and problem-solving ability.
- Curiosity about how complex systems fail.
- Ability to communicate technical findings clearly.
- Evidence of technical initiative through research, open-source projects, CTFs, personal AI projects, coursework, or equivalent work.