1 Data Science

Data Science is the field of collecting, cleaning, analyzing, and interpreting data to find useful information and make decisions.

2 What is Artificial Intelligence?

Artificial Intelligence (AI) is a field of computer science that focuses on creating machines that can perform tasks that normally require human intelligence.

Python vs Java for AI

3 Machine Learning (ML)

Machine Learning is a subset of AI where computers learn patterns from data and improve their performance without being explicitly programmed.

4 Types of Machine Learning

Supervised Machine Learning

The model is trained using labeled data, meaning the correct output is already known. The algorithm is trained on labelled data where the answer is provided.

Unsupervised Machine Learning

The model works with unlabeled data and tries to find patterns or structures in the data. No need to provide trained/labeled data — the model discovers hidden patterns on its own.

Reinforcement Learning

Learning through rewards and penalties. An agent learns by interacting with an environment and receiving rewards or penalties. The agent learns to make decisions by performing actions in an environment to maximize cumulative rewards.

Ensemble Learning

A technique that combines multiple machine learning models to produce a better and more accurate prediction. Combining multiple models for better accuracy.

Neural Network

A computational model inspired by the human brain, consisting of interconnected nodes that process information. It is a brain-inspired model used in deep learning.

Semi-Supervised ML

A type of ML where the model is trained using both labeled and unlabeled data.

5 Deep Learning

Deep Learning is a subset of Machine Learning that uses artificial neural networks with many layers to learn complex patterns from large amounts of data.

6 Generative AI (Gen AI)

Generative AI is a type of AI that can generate new content such as text, images, audio, video, or code based on the data it has learned from training.

In short: Generative AI creates new content instead of only analyzing data.

7 Agent AI

Agent AI refers to autonomous AI systems that can plan, make decisions, and perform tasks automatically using different tools and data sources.

8 Hugging Face

Hugging Face is a company and open-source platform that provides tools, libraries, and models for AI and NLP (Natural Language Processing). It helps developers and researchers easily build, train and use machine learning models, especially for language-related tasks.

9 Embeddings & Vectors

What is an Embedding?

An embedding is a technique or method used to represent data. It is a way of converting data (such as words, sentences, or images) into numerical vectors so that computers can understand and process them.

What is a Vector?

A vector is the numerical output produced by the embedding. It is a list of numbers that captures the meaning or features of the original data.

10 Tokenization

Tokenization is the process of breaking text into smaller units called tokens so a model can process them.

In short: Breaking text into smaller units (tokens).

11 Vectorization

Vectorization is the process of converting data (text, images, audio, etc.) into numerical vectors so that machines can process and understand it.

In short: Converting tokens to numerical vectors.

12 Transformers

Positional Encoding

Positional encoding is a technique used in transformer models to inject information about the order of tokens in a sequence. Since transformers process all tokens simultaneously, positional encoding tells the model where each token appears.

Self-Attention in Transformers

Self-attention allows each word to look at other words in the sentence and decide which ones are important. Words look at each other — giving the model a richer understanding of context.

HNSW — Hierarchical Navigable Small World

HNSW is a graph-based ANN (Approximate Nearest Neighbor) algorithm used to perform fast vector similarity search. It is one of the most popular algorithms for ANN search.

How it works:

1. Start from the top layer
2. Move to nodes closer to the query vector
3. Go down to lower layers
4. Continue until the nearest vector is found

ANN — Approximate Nearest Neighbor

ANN is a technique used to quickly find vectors that are most similar to a query vector, without checking every vector in the dataset. In vector search systems you may have millions or billions of embeddings — computing similarity with every vector would be too slow. ANN finds very close neighbors quickly, but the result may be approximate rather than perfectly exact.

13 Database Types

RDBMS — Relational Databases

A Relational Database Management System stores data in tables with rows and columns. Tables are connected using keys (Primary Keys, Foreign Keys). Uses SQL language for queries.

Examples: MySQL · PostgreSQL · Oracle · Microsoft SQL Server · SQLite · MariaDB · Amazon Aurora

NoSQL Databases

NoSQL means "Not only SQL". A NoSQL database stores unstructured or flexible data and is used for large-scale web applications.

Graph Databases

A Graph Database stores data as nodes and relationships (edges). Used when relationships between data are very important.

Examples: Neo4j · Amazon Neptune · ArangoDB · TigerGraph · OrientDB

14 Vector Databases

A Vector Database stores vectors (numerical embeddings) instead of traditional text data. These databases are mostly used in AI and ML applications.

15 Vector Mathematics for AI Engineers

Cosine Similarity, Euclidean Distance, Dot Product, and L2 Normalization are used in every vector database operation.

1. Cosine Similarity (used for embeddings)

This measures how similar two vectors are by measuring the angle between them.

cos(θ) = (A · B) / (‖A‖ × ‖B‖)

Range: -1 to 1
  1  → very similar
  0  → unrelated
 -1  → opposite

2. Euclidean Distance

This measures the straight-line distance between two vectors in n-dimensional space. Lower = more similar. Uses magnitude (size matters).

d = √Σ(Aᵢ - Bᵢ)²

Small distance  → Similar vectors
Large distance  → Different vectors

3. Dot Product

The dot product multiplies matching elements and adds them. It is the fastest to compute and equals Cosine Similarity when vectors are L2-normalized.

A · B = Σ(Aᵢ × Bᵢ)

Example:
a = [1, 2, 3]  b = [4, 5, 6]
1×4 + 2×5 + 3×6 = 32

16 What is RAG?

RAG = Retrieval-Augmented Generation

RAG is an AI technique where a system retrieves relevant information from external data sources and then uses a generative model to produce an accurate answer.

17 RAG Architecture

Pipeline 1 — Ingestion (Offline / Batch)

Pipeline 2 — Query (Online / Real-time)

18 RAG Challenges

19 What is an AI Agent?

An AI Agent is a system that can perceive its environment, make decisions, and take actions automatically to achieve a goal.

Think of an AI agent like a smart assistant that:

• Observes — what's happening
• Thinks about what to do
• Acts to complete a task

20 Types of AI Agents

How an AI Agent Flows

1. User asks question
2. LLM reads system prompt
3. LLM decides: Answer directly OR use tool
4. Tool runs
5. Result goes back to LLM
6. Final answer generated

21 Multi-Agent Systems (MAS)

Single Agent System

Only one AI agent is doing everything. Example: A coding assistant that reads your request, writes code, and fixes errors — one brain doing all tasks.

Multi-Agent System (MAS)

Multiple AI agents working together, each with a specific role. Think of it like a team:

• Agent 1 → Planner
• Agent 2 → Researcher
• Agent 3 → Executor
• Agent 4 → Reviewer

All cooperate to finish a task.

Building a Multi-Agent System

Example: Take a topic and create a short blog post.

1. Planner Agent → decides structure
2. Writer Agent → writes content
3. Reviewer Agent → improves it

# Step 1: Choose Tools
# Popular choices: LangChain · AutoGPT · CrewAI

# Step 2: Install requirements
pip install crewai openai

# Step 3: Define Agents
# Step 4: Define Tasks (tasks tell agents what to do)
# Step 5: Create the Crew (connect agents)

# Step 6: How it works (flow)
# 1. Planner  → creates outline
# 2. Writer   → uses outline
# 3. Reviewer → improves final output

22 Orchestration System

An Orchestration System manages and coordinates multiple agents, tools, and steps to complete a task.

In AI systems, an orchestration system controls:

1. Which agent runs next
2. How data flows (state)
3. When to loop
4. When to stop
5. When to ask human

23 GCP & Cloud Computing

What is Cloud Computing?

Cloud computing means using remote servers on the internet instead of local computers. Instead of saving data on your laptop, you store it in the cloud.

What is GCP?

GCP = Google Cloud Platform. GCP is a cloud computing platform provided by Google that offers services like computing power, storage, databases, machine learning, and networking over the internet.

GCP allows developers and companies to run applications, store data and build software on Google's servers instead of their own computers.

24 FastAPI for AI

FastAPI is a modern Python web framework used to build high-performance APIs. In AI projects it is the standard tool for exposing ML models as REST APIs.

# Step 1: Create project folder
mkdir project && cd project

# Step 2: Create virtual environment
python -m venv venv
venv\Scripts\activate

# Step 3: Install required libraries
pip install fastapi uvicorn pytest

# Step 4: Create API file
notepad sentiment-api.py

# Step 5: Run the API
uvicorn sentiment-api:app --reload
# Visit: http://127.0.0.1:8000/docs

# Step 6: Test with Postman
# Step 7: Create unit test file
notepad test-api.py
pytest

25 Streamlit for UI

Streamlit is a Python framework used to quickly build web-based user interfaces (UI) for data science, machine learning, and AI applications.

pip install streamlit

# app.py
import streamlit as st

st.title("My AI App")
user_input = st.text_input("Ask me anything:")
if user_input:
    st.write("You asked:", user_input)

# Run
streamlit run app.py