Learn how to think, code, and experiment like engineers at OpenAI, DeepMind, and Anthropic.
Presale Discount ends Jan 18
Many directions, unified fundamentals
Stop being a consumer of AI. Start becoming a contributor. Master the first principles that drive the cutting edge of research.
Master the fundamental operations that power AI: Dot Products, Softmax, and Tensor Broadcasting.
Understand Autograd, Computational Graphs, and Backpropagation by building your own engine from zero.
From Bigrams to deep MLP architectures. Master Tokenization (BPE) and stability techniques.
Deep dive into Self-Attention, RoPE, and the architecture that changed the world.
Diffusion and VAEs. Implement UNets and VAEs from first principles.
Train agents using Policy Gradients and DQN. Understand Rewards and Environments.
How to read papers, formulate hypotheses, and run high-signal experiments.
"You only need part of this course to start writing your own AI research papers. Follow your curiosity."
We've moved our core community and premium research content to Skool. Join 19K+ scholars building the future of AI.
Get Access NowVuk Rosić
I am an AI Research Scientist specializing in Transformers, LLMs, Deep Learning, and Reinforcement Learning. With over 500+ videos and a community of 19.5K+ subscribers, I focus on breaking down the most advanced AI concepts through first principles.
My goal is to help you stop being just a consumer of AI and start becoming a contributor. This curriculum is designed to equip you with the technical and mathematical intuition needed to read the latest papers from labs like OpenAI or DeepMind and begin publishing your own original research.
Aspiring AI researchers who want to join OpenAI, DeepMind & other top labs, engineers looking to move into research labs, and curious builders who want to understand how to do AI research from the ground up.
Yes. In fact, you only need about 60% of this course to start writing your own AI research papers. We focus on the methodology: how to read papers, formulate hypotheses, and run efficient experiments, so you can start contributing as soon as possible.
Basic Python is helpful, but we cover the mathematics and PyTorch fundamentals from scratch.