Director, Model Post-Training and Agentic Research (AI)

TL;DR

Director, Model Post-Training and Agentic Research (AI): Building post-training and reinforcement learning capabilities for security-specialized AI systems with an accent on RLHF, RLAIF, and agentic research. Focus on designing reward models, building agent harnesses for complex cyber workflows, and developing evaluation methodologies for end-to-end task completion.

Location: Remote (Must be based in the USA)

Salary: $195,000 - $290,000 per year

Company

Global leader in cybersecurity protecting people, processes, and technologies via an AI-native platform.

What you will do

• Own and personally drive the full post-training pipeline, including SFT, RLHF/RLAIF, agent-RL, and reward modeling.
• Build agent-RL training environments and harnesses, focusing on scaffolding, tool-calling interfaces, and planning loops.
• Develop evaluation methodologies for the full agentic stack to measure tool-use reliability and planning coherence.
• Partner with internal teams to integrate post-training and agentic work into the broader model development loop.
• Recruit and lead a high-density team of research scientists and ML engineers, setting the technical bar through active contribution.

Requirements

• MS or PhD in Computer Science, Machine Learning, or a related quantitative discipline.
• 8+ years of experience in ML research or engineering with meaningful depth in LLM post-training.
• Hands-on expertise in SFT data pipelines, RLHF/RLAIF, PPO, and reward model design.
• Demonstrated experience building agentic system harnesses, including tool-use frameworks and memory management.
• Track record of running high-velocity research programs with disciplined tracking and iteration.
• Must have US work authorization (E-Verify participant).

Nice to have

• Experience building RL training environments, including rollout infrastructure and reward shaping.
• Applying RL techniques in security or adversarial ML domains.
• Published research in post-training, RLHF, or RL for language agents at top-tier venues (NeurIPS, ICML, ICLR, ACL).
• Experience adapting open-weight base models (Llama, Qwen) for domain-specialized training.
• Familiarity with security practitioner workflows, such as penetration testing or incident response.

Culture & Benefits

• Market leader in compensation and equity awards.
• Comprehensive physical and mental wellness programs.
• Competitive vacation and holidays for recharge.
• Paid parental and adoption leaves.
• Professional development opportunities for all employees regardless of level.