Hello again 👋🏾
I recently took a career break to reposition. I wasn’t burned out. I just knew it was time to go deeper.
Artificial Intelligence (AI) - how machines learn, make decisions, and interact like humans.
AI isn’t just one thing. It’s a big umbrella with several moving parts. So, I will walk you through:
- Why I chose AI
- The building blocks of AI explained in simple terms
- How I am approaching it as a beginner
🤔 Why AI?
I started paying attention when I saw tools like AI Customer Support, Notion AI, Copilot, and other assistants simplifying life in big ways.
These tools made me ask:
- How do they understand language so well?
- Can I build something like this?
- What skills do I need to augment AI and web development?
That’s when I began learning.
🧱 The Building Blocks of AI
Here are the key layers I have been learning, one block at a time:
1. Data
AI systems are data-hungry. Just like a child learns from experience, machines learn from examples.
- For a machine to recognize cats, you feed it thousands of cat images.
- For it to respond to human questions, it needs millions of text samples.
👉🏽 Data is the foundation. No good data, no good AI.
2. Machine Learning (ML)
This is where the learning happens.
Instead of hardcoding rules, we train algorithms to learn patterns from data.
- Show the algorithm enough spam emails versus real emails, and it will learn to detect spam emails on its own.
- It’s not magic, just math 🦸🏼♂️.
- Tools like Scikit-learn and XGBoost are used to build these models
There are 3 types of ML:
-
Supervised Learning – we give the machine labeled data to learn to identify the labels of new/unseen data. Example is a list of emails with each labelled as
spamornot_spam. When a new email is fed it, it identifies ifspamornot_spam. - Unsupervised Learning – we give the machine unlabelled data and it finds patterns by itself. Example is a thousands of pictures of cat and dogs without labelling each of them. The machine groups the pictures based on similarities. When a new picture is fed it, it identifies the group it belongs to.
- Reinforcement Learning – it learns by trial and error just like teaching a robot to walk.
3. Deep Learning
This is a subset of ML, but more powerful and inspired by how the human brain works.
It uses Neural Networks with layers of neurons that process data in steps.
- This is how we get voice recognition, image detection, and language generation.
- It’s the engine behind tools like ChatGPT, Siri, and face recognition apps.
Deep learning needs:
- Big data
- Strong computing power (GPUs)
- Frameworks like TensorFlow and PyTorch
4. Computer Vision
This is a subset of DL that allows machines to see and interpret visual information.
- Face detection on your phone
- Object detection in security systems
- Medical image diagnosis
5. Natural Language Processing (NLP)
This is the part of AI I fell in love with 🥳.
NLP is a subset of DL and it is how machines understand and work with human language.
- Translating from English to French
- Summarizing a blog post
- Answering your questions like a chatbot
Key ideas in NLP include:
- Tokenization - breaking sentences into words
- Embeddings - turning words into numbers
- Transformers - the magic architecture behind most LLMs like GPT
6. Large Language Models (LLMs)
This is where things get really exciting 🤠.
LLMs are subsets of NLP. They are trained on huge datasets of text to understand and generate human-like responses.
Popular LLMs include:
- GPT by OpenAI
- BERT by Google
- LLaMA by Meta
They can:
- Answer questions
- Write code
- Explain concepts
- Help with customer service
- And even power search, recommendations, or writing assistants in your web apps
I am currently learning:
- How to fine-tune LLMs for specific tasks
- How to integrate them into web applications
- How to build smart tools with them
📌 The 3 Types of AI (Based on Capability)
| Type | Description | Status |
|---|---|---|
| Narrow AI (Weak AI) | AI that does one thing very well (e.g., ChatGPT, Google Translate) | Already here |
| General AI (Strong AI) | AI that can think and reason like a human | Still a research goal |
| Super AI | AI smarter than humans in every aspect | Purely theoretical |
📌 Typical AI Workflow
Collect Data 👉 Clean and label it
Train Models 👉 Use algorithms to teach the machine
Test and Use Models 👉 Make predictions or decisions in real-world scenarios
🔍 My Learning Approach
I didn’t try to learn everything at once. I broke it into stages:
- ✅ Understand theory: ML basics, math, workflows
- ✅ Learn tools: Python, NumPy, Pandas, Matplotlib, Scikit-learn, TensorFlow
- ✅ Learn the Data Science workflow.
- ✅ Practiced building ML and DL models: mental health predictor, image classifier, text classifier
- ✅ Create projects on LLMs: chatbot, question and answer system from private data
- ✅ Share learnings: write and teach
✨ Why I am Sharing This
I believe in teaching while learning, and this post is part of a new direction in my career. I want to become a Backend Engineer with AI-powered system experience, focused on blending LLMs, NLP, and Web Development to build more intelligent, inclusive, and helpful systems.
So, this is me:
- Learning from scratch
- Building intelligent tools
- Teaching as I go
- And open to opportunities, collaboration, and contribution
💡 What’s Next?
I will go deeper into:
- Machine Learning
- Deep Learning
- NLP and LLMs
- And how all this ties into real-life web development
- And later share my roadmap to blending Software Engineering and AI Engineering
If this helped you, follow along. I will keep it simple, honest, and useful.
Happy coding!
Top comments (0)