Inside AI Models

Roadmap

The AI learning roadmap

Everything an AI researcher stands on — and how each topic builds on the last. Start at the foundations, climb to the top.

Goal

AI Researcher

Reads, reproduces, and pushes the state of the art — combining every layer below.

Builds on: Fine-tuning & RLHF, RAG & Retrieval, Agentic Workflows

Building with LLMs

Fine-tuning & RLHF

Adapting pretrained models — SFT, LoRA, preference tuning.

Builds on: Transformers & LLMs

RAG & Retrieval

Grounding models in external knowledge with retrieval.

Builds on: Transformers & LLMs

Agentic Workflows

LLMs that plan, use tools, and act in multi-step loops.

Builds on: Transformers & LLMs

Language Models

Transformers & LLMs

Attention, scaling, pretraining — the modern backbone.

Builds on: Deep Learning

Read the article →

Deep Learning

Deep Learning

Neural networks, backpropagation, training at scale.

Builds on: Optimization, Core ML

Core ML

Optimization

Gradient descent and how models actually learn.

Builds on: Linear Algebra, Calculus

Read the article →

Core ML

Regression, classification, generalization, evaluation.

Builds on: Probability, Statistics, Programming, Optimization

Math foundations

Linear Algebra

Vectors, matrices, tensors — the language of data.

Read the article →

Calculus

Derivatives and gradients that drive learning.

Probability

Uncertainty, distributions, likelihood.

Statistics

Estimation, inference, hypothesis testing.

Programming

Python, NumPy, PyTorch — making it run.