Junior Software Engineer

The role

You will design, build, and refine RL tasks. You will own the full lifecycle of tasks including idea generation, implementing grading infrastructure, running models, and analyzing failures.

What you will do

• Design, build, and refine RL tasks
• Come up with task ideas
• Implement grading infrastructure
• Run frontier models against tasks
• Analyze where and why models fail
• Iterate until tasks are rigorous and fair
• Use coding agents heavily
• Direct coding agents and evaluate their output

Required qualifications

• Have just graduated or are about to graduate
• Can code in Python
• Are confident working independently
• Are motivated by problems that require both technical skill and creative cleverness

Nice to have

• Candidates who have spent significant time working with frontier coding agents

Benefits

• health
• dental
• vision
• life insurance

About the company

Mechanize builds reinforcement learning environments that frontier AI labs use to train and evaluate their coding models. It is a ~20 person team in San Francisco.