Courses AI Foundations In collaboration with Google DeepMind
Google DeepMind: AI Research Foundations
A free, university-level AI curriculum from Google DeepMind, developed with University College London
About this path
AI Research Foundations is an eight-course curriculum that introduces learners to the foundational building blocks of modern AI through the lens of language modeling. It is designed to move past AI literacy and give technical learners a deep, applied understanding of how systems like Google Gemini are actually built, starting from tokenization and neural networks, then moving through transformer architectures, fine-tuning, and alignment.
Research skills and responsible innovation are central to every course, so learners are equipped to build high-impact AI solutions globally. The curriculum builds up to a capstone project where learners research and develop an AI-backed solution to a challenge they care about.
It is designed for university students and community learners who are proficient in Python and studying computer science, mathematics, physics, or other technical subjects.
The eight courses
Released sequentially on Google Skills:
- Build Your Own Small Language Model. Fundamentals of language models and ML development pipelines.
- Represent Your Language Data. Preparing text data for language models to process.
- Design and Train Neural Networks.
- Discover the Transformer Architecture.
- Finetune Your Model.
- Align Your Model.
- Accelerate Your Model. Scaling training with GPU resources, managing compute and memory.
- Capstone: Develop Your Model for Real-World Impact. Responsible AI content runs throughout, including ethical dataset design, stakeholder mapping, and environmental considerations such as energy use.
Format
- Online and self-paced, or teachable in blended mode at a university, college, or grassroots meet-up
- Each course combines reading, videos, practical labs, and short assessments
- Hands-on labs run in the Google Cloud console
- Skill badges validate practical knowledge through challenge-based assessments
- Free on Google Skills, with a monthly credit allocation for lab access
Prerequisites
- Proficiency in Python
- A background in a technical subject (CS, math, physics, or similar)
- No prior machine-learning experience required
Who it’s for
University students and community learners who want a serious, research-oriented introduction to how modern language models actually work. Not just how to call an API, but how the systems behind them are designed, built, trained, fine-tuned, aligned, and scaled.
Open-source materials
See full course information here: https://www.skills.google/paths/3135 The practical labs are openly available at github.com/google-deepmind/ai-foundations. Software is under Apache 2.0, course materials under CC-BY 4.0.
Developed in collaboration with pedagogy experts and academics at University College London. Google.org is also supporting the curriculum’s rollout in African classrooms through lecturer training and instructional toolkits.