AI Builder

Goal: Understand what AI is, how modern AI learns, the main AI problem types, how models are used through inference, and where AI fits into products.

1. Day01

   Explain AI in simple terms and distinguish AI systems from ordinary software.

2. Day02

   Explain the basic idea of learning from data rather than hand-coding every rule.

3. Day03

   Identify common AI problem types and match a product problem to an appropriate AI approach.

4. Day04

   Explain what a model does and how inference turns an input into an output.

5. Day05

   Translate a real product requirement into an AI-enabled feature and explain where the model fits.

Goal: Build a practical mental model of neural networks, training, loss, gradients, deep learning, NLP and attention.

1. Day06

   Explain a neural network as layers that transform inputs into useful representations.

2. Day07

   Explain how loss measures error and how gradients guide model improvement.

3. Day08

   Explain why deeper networks can learn richer representations and where deep learning is useful.

4. Day09

   Explain how text can be represented and processed for language tasks.

5. Day10

   Explain attention as a way for a model to focus on relevant parts of an input.

Goal: Understand Transformers, LLMs, tokens, context, generation and model selection.

1. Day11

   Explain the high-level Transformer pipeline and why it changed language AI.

2. Day12

   Explain what an LLM does and what kinds of tasks it can perform.

3. Day13

   Explain tokens, context windows and why context affects an LLM's output.

4. Day14

   Use an LLM API and explain the basic request → model → response flow.

5. Day15

   Choose an appropriate model for a simple task using capability, quality, latency and cost considerations.

Goal: Turn LLM knowledge into usable application features through prompts, structured outputs and task decomposition.

1. Day16

   Build a simple application that sends user input to an LLM and uses the response.

2. Day17

   Write prompts with clear instructions, context, constraints and expected output.

3. Day18

   Use zero-shot and few-shot examples and explain when examples improve results.

4. Day19

   Make an LLM return predictable structured data that an application can consume.

5. Day20

   Break a larger AI task into smaller model calls or deterministic steps.

Goal: Learn to evaluate prompts and use embeddings and vector search to improve relevance.

1. Day21

   Compare prompt variants using a repeatable test set and simple quality criteria.

2. Day22

   Explain embeddings as numerical representations of semantic meaning and generate them for text.

3. Day23

   Retrieve semantically similar content using vector search.

4. Day24

   Define the problem, users, inputs, outputs and AI workflow for Mini Project 1.

5. Day25

   Build the first working version of Mini Project 1 and test it with representative inputs.

Goal: Understand grounding, RAG and RAG quality, and begin building agents.

1. Day26

   Improve Mini Project 1 from testing feedback and make a meaningful functional change.

2. Day27

   Identify common hallucinations and explain how grounding can reduce unsupported answers.

3. Day28

   Build a basic retrieval-augmented generation flow: retrieve relevant context, then generate an answer.

4. Day29

   Test retrieval and answer quality and identify where a RAG pipeline fails.

5. Day30

   Explain an agent as a system that can reason through a task and use available actions/tools.

Goal: Build tool-using workflows and understand memory, vision/OCR and voice AI.

1. Day31

   Connect an LLM to a tool and explain the input → tool → result → response loop.

2. Day32

   Build a simple multi-step workflow and explain what information should persist between steps.

3. Day33

   Use AI to extract or reason over information in images and documents.

4. Day34

   Explain the speech-to-text → reasoning → text-to-speech flow and prototype a voice interaction.

5. Day35

   Define the problem, workflow, AI capabilities and success criteria for Mini Project 2.

Goal: Finish an AI application and make it safer, testable and ready to explain.

1. Day36

   Build and test the core workflow of Mini Project 2.

2. Day37

   Polish the project, fix failures and improve the user experience.

3. Day38

   Identify common AI risks and add basic guardrails for unsafe or invalid inputs/outputs.

4. Day39

   Define practical tests and explain the production concerns that matter for an AI application.

5. Day40

   Design the capstone AI system and explain its components, flow, trade-offs and next steps.