AIML - Machine Learning Manager, Red Teaming - Responsible AI and Safety

Written by the T&S Network from Apple's posting.

Leads a red-teaming function inside Apple's Responsible AI and Safety team that actively probes Apple Intelligence features for safety and security risks before and after they ship. The team sets its own agenda, separate from standard evaluation, to find real vulnerabilities in generative AI systems rather than measure known ones.

• AI misuse and jailbreaks

What you'd do

• Hire and grow red-teaming engineers who each take long-term ownership of a set of features
• Set agenda focused on active vulnerability discovery, kept separate from standard evaluation
• Set coverage for pre-ship checks, post-launch monitoring, and foundation model testing alongside model teams
• Set the split of duties with neighboring security teams on shared risks like prompt injection
• Run outside red-teaming vendors, weighing their worth, price, and how they plug into the pipeline
• Set measures of program success and brief leadership on them regularly
• Guide the technical design of automated red-teaming tools

What it asks for

• Master's or PhD in computer science, ML, statistics, or a related field, or equivalent background
• Hands-on time building or testing generative models for evaluation or products
• Strong production-quality Python engineering skills
• Deep experience in foundation model-based programming, such as DSPy for prompt optimization
• Practice working with messy labels from crowd workers and human raters
• Publication record at conferences such as NeurIPS, ICML, ICLR, or EMNLP (preferred)
• Experience on Responsible AI or AI safety work (preferred)

Worth knowing

• Works with highly sensitive content, including exposure to offensive and controversial material

Terms in this role

Red teaming.

Testing a system, often an AI model, by deliberately trying to make it fail or misbehave, so problems are found and fixed first.

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on Applenorth_east before you apply. If we got something wrong, report this listing.