Researcher, Pretraining Safety (AI)

TL;DR

Researcher, Pretraining Safety (AI): Develop and evaluate safety mechanisms in early-stage AI models focusing on pretraining architectures and safe-by-design approaches with an accent on identifying unsafe behaviors early, designing safer architectures, and improving controllability. Focus on designing novel safety evaluation techniques, data curation strategies, and collaborating across safety teams to reduce risks before deployment.

Location: Onsite in San Francisco, United States

Salary: $310,000–$460,000 + Equity

Company

hirify.global is an AI research and deployment company dedicated to ensuring that general-purpose artificial intelligence benefits all of humanity by pushing AI capabilities and deploying them safely.

What you will do

• Develop techniques to predict, measure, and evaluate unsafe behavior in early-stage models
• Design data curation strategies to improve pretraining priors and reduce downstream risk
• Explore safe-by-design architectures and training configurations to improve controllability
• Introduce novel safety-oriented loss functions, metrics, and evaluations into the pretraining stack
• Collaborate with cross-functional safety teams to unify pre- and post-training risk reduction

Requirements

• Must be located onsite in San Francisco, United States
• Experience developing or scaling pretraining architectures such as LLMs, diffusion, or multimodal models
• Proficiency with training infrastructure, data pipelines, and evaluation frameworks (Python, PyTorch/JAX, Apache Beam)
• Strong data-driven approach with statistical reasoning and experimental design rigor
• Ability to collaborate with diverse technical and cross-functional partners

Culture & Benefits

• Equal opportunity employer with commitment to diversity and inclusion
• Competitive compensation including salary and equity
• Work in a mission-driven AI safety focused environment
• Access to reasonable accommodations for applicants with disabilities