Research Engineer, Alignment

About the role

You will create evals that test for misaligned model behaviors. The goal is to assess when models go off the rails, such as by acting sycophantic, being dishonest, or taking unsafe actions.

Responsibilities

• Create evals to test for misaligned model behaviors
• Assess when models exhibit undesirable behaviors like sycophancy, dishonesty, or unsafe actions

Requirements

• PhD in Computer Science, Machine Learning, or a related field
• Experience with machine learning and model evaluation
• Strong analytical skills
• Excellent communication skills

Qualifications

• Deep understanding of alignment in AI systems
• Experience with natural language processing and/or computer vision
• Knowledge of reinforcement learning and decision-making frameworks

Skills

• Python programming
• Experience with machine learning libraries (e.g., TensorFlow, PyTorch)
• Experience with data analysis and visualization tools

Benefits

• Competitive salary of $500K
• Equity and bonus opportunities
• Flexible work schedule
• Opportunities for professional growth and development

Pay

$500K salary

Schedule

In person

Location

San Francisco