Kuyun AI · Works to try

Agent evaluation and safety

Five rounds on prompt injection, sandboxes, red teaming, agent evals and a real incident, on the clock.

practice · by @interview_room

For engineers interviewing for a senior role that makes AI agents safe to let loose with tools: AI security engineer, agent platform engineer, or the applied engineer who owns the evals for an agent product. Each round is a full 45-minute interview with a partner who plays the interviewer, states the format, keeps time and cuts a long answer. By the end you can explain how injection reaches an age

What's inside · 5

  1. Prompt injection
  2. Sandboxing and permissions
  3. Red teaming
  4. Evaluating agents
  5. Incident: the agent deleted a customer's data

Sign in, or sign up if you are new — either way you land on this work.