Head of Frontier AI Research — Lead RL Team & Roadmap

Ghost AI; New York (NY); Híbrido; USD 300.000 - 600.000; Jornada completa

Descripción de la vacante

Ghost AI is seeking a hands-on Head of Research to own the roadmap and guide frontier AI initiatives from discovery to deployment. You will lead a growing team of researchers, tackle post-training challenges, and collaborate with top labs to push the boundaries of RL and model training.

You will stay hands-on with experiments, bridge research and engineering, and shape the company's technical direction as it scales its deterministic RL environments and datasets.

Formación

• 8+ years of experience as a research scientist or applied researcher.
• Proven leadership experience at a frontier AI lab, leading AI research lab or AI data company.
• Deep, hands-on expertise in reinforcement learning, model training and post-training.
• A track record of impactful research — a notable publication or model that has performed strongly on a recognised benchmark.
• Experience setting research direction, leading projects and delivering measurable results.

Responsabilidades

• Own the research roadmap and drive the strategy for frontier AI initiatives.
• Tackle challenging post-training problems with leading AI research teams.
• Build an applied research function from the ground up and scale the team.
• Bridge research and engineering to turn breakthroughs into production-ready deliverables.

Conocimientos

Reinforcement Learning

Model Training

Post-training

Research Leadership

Experimentation

Educación

PhD in RL-related field

Herramientas

GRPO

Qwen 2.5

Descripción del empleo

Ghost AI is seeking a hands-on Head of Research to own the roadmap and guide frontier AI initiatives from discovery to deployment. You will lead a growing team of researchers, tackle post-training challenges, and collaborate with top labs to push the boundaries of RL and model training.

You will stay hands-on with experiments, bridge research and engineering, and shape the company's technical direction as it scales its deterministic RL environments and datasets.