Research Lead, Pre-Training Safety

• In this role, you'll lead pre-training safety research, focusing on removing harmful capabilities from models at their source.
• Build and lead a team to scale methods like Deep Ignorance to frontier-scale models using data filtering and gradient routing.
• Conduct novel research to isolate and remove dual-use capabilities from model components and reduce loss-of-control risks.
• Partner with red teams to stress-test models and analyze how safety methods scale to superhuman AI systems.

FAR AI aims to ensure AI systems are trustworthy and beneficial to society. They incubate and accelerate research agendas that are too resource-intensive for academia but not yet ready for commercialisation by industry.