What you'll learn

Build a GPT-style large language model from scratch in Python and PyTorch, without ready-made transformer layers

Explain self-attention, Query/Key/Value, the causal mask and multi-head attention with clear visual intuition

Code tokenization (character-level and Byte-Pair Encoding), embeddings and positional encoding by hand

Implement a complete transformer block with layer normalisation, residual connections and a feed-forward network

Train a transformer on a laptop CPU and read loss curves to diagnose learning and overfitting

Generate text with temperature, top-k and top-p sampling, and visualise next-token probabilities and attention

Build an interactive AI playground web app with Gradio to demo your own language model

Explain how GPT, ChatGPT, BERT and the original Transformer relate, including pre-training, fine-tuning and RLHF

Answer common transformer interview questions with confidence

Requirements

Basic Python programming (functions, classes, lists, loops)

A computer with Python 3.9 or newer. No GPU required

No prior deep learning or advanced maths needed. Every concept is explained visually first

Description

This course contains the use of artificial intelligence.Ever wondered what really happens inside ChatGPT? In this course, you'll find out by building one yourself.Build Your Own ChatGPT takes you from zero to a working GPT-style language model, written in Python. No black boxes and no shortcuts. You'll write every line and understand why it works.What you'll build, step by step:A tokenizer: use byte-pair encoding to turn raw text into tokens and numbersEmbeddings: represent tokens as vectors the model can learn fromSelf-attention: queries, keys, values, causal masking and multi-head attentionThe transformer: feed-forward layers, residual connections and layer normalizationTraining: train your model on real text and watch the loss fallText generation: sample from your model and chat with your own GPTEvery concept is explained visually first and then coded from scratch, so the math becomes clear and the code makes sense.Who this course is for:Developers who use LLMs and want to understand how they really workStudents and data scientists moving into AI and NLPAnyone curious about the technology behind ChatGPTRequirements: basic Python. Familiarity with high school math helps, and everything else is taught along the way.By the end, you won't just use AI. You'll know how to build it.Enroll now and start building your own ChatGPT today.

Python developers who want to understand how ChatGPT and large language models really work,Students and data scientists moving into Generative AI and deep learning,Software engineers preparing for machine learning or AI engineering interviews,Professionals who use LLM tools or APIs and want to see what happens inside the black box,Educators and tech leads who need to explain transformers clearly to their teams