Technical AI safety is the work of making sure advanced AI systems behave as intended: testing what they can do, finding how they fail, understanding how they work inside, and developing ways to keep them under meaningful human control. The field is growing fast, and it needs more people from more places, including Africa, where AI systems are used in many languages and contexts that today’s safety research rarely covers.

This hands-on course helps you get into the field by building a public portfolio: evaluation tasks, an interpretability mini-project, a paper reproduction and written research, plus a contribution to the AI safety community. By the end, you’ll have evidence of your skills that you can share when exploring fellowships, research programmes, jobs and collaborations.

Who it’s for

Students, software developers, data scientists, researchers and career-changers who want to explore technical AI safety. You should be comfortable with basic Python. No prior machine learning research experience is needed; the course explains the concepts you need and points to further learning.

What you’ll be able to do

• Explain the main areas of technical AI safety and the risks they address.
• Describe how modern AI models are built and trained, and where safety problems arise.
• Design, build and publish evaluation tasks using open-source tools.
• Carry out basic red-teaming and test safety across languages.
• Run a small interpretability experiment.
• Read, reproduce and write up research.
• Explore opportunities and prepare strong applications.

How it works

Eleven modules, about 48 hours, at your own pace. Short lessons, labs with optional code, and practice activities. Assessment includes module quizzes, four portfolio assignments, a final exam and a capstone public portfolio.

An important note

PACFAIE is not affiliated with any AI lab, fellowship, research programme or employer mentioned in this course, and completing it does not guarantee admission to any programme or job. We mention organisations and tools only as examples to help you explore the field. Always check each programme’s own website for current requirements and deadlines.

Certificate

Complete every step to earn the Certificate in Technical AI Safety Foundations, a verifiable digital certificate and Open Badge, showing your capstone result.