Director, AI Security Science (Remote)

The frontier of cybersecurity is being reimagined by AI, and the defenders who win the next decade will be the ones who leverage agents that actually work in the constraints of the environment. We’re looking for an applied research lead to take on both sides (offensive and defensive) of this frontier and help turn it into products.

In this role, you’ll set the scientific direction for our agentic AI work and lead the research behind the next generation of security products. The problems span the full stack: post-training small, efficient models; designing and tuning the harnesses; and developing the evaluation and verification methodologies that give us rigor around if the agents are actually doing what we built them to do. These are open research problems, and we expect the person in this seat to advance the state of the art on them.

This is a player/coach role in the truest sense. You’ll lead teams of strong scientists and engineers, but you’ll also be in the work itself. The successful candidate will bring technical credibility, instincts sharpened by doing the work, and leads by raising the bar on what the team believes is possible.

What You’ll Do:

• Lead the research agenda for agentic AI systems that operate autonomously across cybersecurity use cases. Set scientific direction, shaping the team’s research bets and staying close enough to the work to run experiments, review results, and dig into hard problems alongside the team.

• Drive original research on efficient models tuned for security workflows, including methods in post-training, reward modeling, and data curation strategies that enable them. Personally lead the experimental work on the most ambiguous problems.

• Establish a rigorous evaluation, verification, and validation methodology covering models, data, agent designs, and end-to-end systems. This includes current and new benchmarks, adversarial evaluation protocols, and measurement practices that others in the field will adopt.

• Partner closely with security researchers, threat intel, detection engineering, and product to ground applied research in real operator workflows and adversarial conditions.

• Define principled abstractions for the skills and tools agents use, grounded in practical ways that security agents execute.

• Advance the science of agent design, including harnesses, scaffolding, planning, memory, tool use, and multi-step reasoning. Contribute novel techniques back to the field through publications and standards.

What You’ll Need:

• MS or PhD in computer science, computer engineering, or similar quantitative field.

• 10+ years of experience in data science, ML/DL, or AI with a focus on large-scale data problems.

• Hands-on expertise in modern LLM post-training techniques.

• Demonstrated research contributions to or experience building agentic systems.

• Strong evaluation discipline (designing, extending, or validating methodologies).

• A leadership style that values hands-on contributions at every level, including your own.

Ways to Stand Out:

• Hands-on background in offensive security, detection engineering, incident response, or threat intelligence applied research.

• Experience designing RL environments and synthetic data pipelines for agent tuning.

• Experience navigating regulatory, compliance, or audit conversations about AI systems, especially translating research into responsible deployment.

• History of mentoring and growing researchers in the security domain.

LI-Remote
LI-RC1