Learn how AI thinks by stepping
inside a neural network you can touch.
A small learning studio for absolute beginners. No prior code, no abstract math walls — you'll move from a single neuron to your first image classifier, one activation at a time.
- ~24h
- guided sessions
- 9
- core concepts
- 3 wks
- since launch
- 0 → 1
- your first model
Modules emerge before introductions.
We won't bury the course under a brand story. Here are the nine knowledge nodes you'll travel through, in roughly the order their signals fire.
-
node · 01 concept
What is Artificial Intelligence?
Where 'AI' actually lives in software you already use — and where it doesn't.
weight 0.92 · bias 0.38 -
node · 02 mental model
Neural Networks, gently
Neurons, layers, signals. A working mental model before any code.
weight 0.88 · bias 0.31 -
node · 03 build
The Perceptron
The smallest decision-maker. We build one by hand.
weight 0.99 · bias 0.24 -
node · 04 data
Training Data
Why data shape matters more than model size when you're starting.
weight 0.42 · bias 0.12 -
node · 05 math, lightly
Loss Functions
How a network knows it's wrong — and what it does about it.
weight 0.48 · bias 0.09 -
node · 06 build
Your First Simple Model
End-to-end: data in, prediction out, in a single notebook.
weight 0.85 · bias 0.18 -
node · 07 vision
Image Classification Basics
Pixels → features → a model that can tell cats from not-cats.
weight 0.76 · bias 0.22 -
node · 08 language
Text Examples
Tokens, embeddings, and tiny language tasks you can run yourself.
weight 0.63 · bias 0.15 -
node · 09 lab
Google Colab Exercises
Free GPU notebooks. No setup. Just open and learn.
weight 0.57 · bias 0.33
The learning roadmap, traced as a signal.
Sense the shape of AI
Build intuition. Watch a perceptron learn AND, OR, then XOR fail spectacularly.
Data, loss, gradient
Three words that sound scary and aren't. We meet each one with a slider, not a chalkboard.
Your first real model
A small image classifier in Colab. You'll watch its loss curve breathe.
Pick a path
Vision, text, or a tiny personal project — guided, not graded.
How a neural network actually works.
node: //neuronsA neuron takes numbers, multiplies them by weights, adds a bias, and squashes the result. That's it. Everything else is repetition.
A layer is many neurons in parallel. A network is layers stacked so each one reads the previous one's opinion.
Training is the loop: predict, measure how wrong, nudge every weight a little in a better direction, repeat — often millions of times.
predict → loss → gradient → nudge → predict …
AI Concepts Playground.
Drag the sliders. This single neuron fires only when its inputs and weights agree. Same arithmetic that powers the largest models alive — just smaller.
y = σ(z) = 0.500
layer 05 · activations[]
Student projects, as fired nodes.
These aren't case studies. They're notebooks beginners actually finished. Each one is a node that lit up after the right inputs.
-
PJ-01 activated
Cat vs. Not-Cat
1-layer model on a 200-image set. Accuracy curve drawn live in Colab.
acc 0.86open ↗ -
PJ-02 activated
Tip-Sentiment
Tiny logistic model classifying restaurant reviews as 👍 or 👎.
f1 0.78open ↗ -
PJ-03 activated
Hand-drawn Digits
MNIST with a 2-layer net, end-to-end in one notebook.
acc 0.94open ↗ -
PJ-04 activated
XOR, finally
Why one neuron can't do it, and what changes when you add a hidden layer.
loss ↓open ↗ -
PJ-05 activated
Spam Sieve
Text → bag-of-words → tiny classifier. Watch the false positives.
prec 0.81open ↗ -
PJ-06 activated
Pixel Painter
Generate a 16×16 image from a noise vector. Your first generative toy.
epoch 50open ↗
Outputs from real learners.
We're three weeks old. These are early beta students whose models successfully converged — and whose confidence did too.
-
PJ-03 → Ana D. weight +0.92
"I built a digit recognizer the second weekend. Not a tutorial copy — I actually understood the layers."
-
PJ-01 → Marcus L. weight +0.88
"First time math felt like a tool, not a wall. The perceptron page broke it open for me."
-
PJ-05 → Reem K. weight +0.81
"Plain language, no hype. Three weeks in, I shipped a spam filter on my own emails."
-
PJ-06 → Iván R. weight +0.79
"I'd bounced off two huge AI courses before this. The pacing here is humane."
A typical student trajectory.
layer 08 · historical context
An interactive AI timeline.
-
1943McCulloch–Pitts neuron
The first mathematical idea of a neuron, on paper.
-
1958Perceptron
Rosenblatt builds a learning machine — yes, the one you used above.
-
1986Backpropagation
Networks finally learn deep representations efficiently.
-
2012AlexNet
Deep learning wins ImageNet. Vision changes overnight.
-
2017Transformers
Attention reshapes language modeling — the seed of modern LLMs.
-
NowYou
An interactive moment where understanding becomes accessible.
What a Colab session actually looks like.
# Step 6: train a tiny image classifier
import torch, torch.nn as nn
from torchvision import datasets, transforms
train = datasets.MNIST(".", download=True, train=True, transform=transforms.ToTensor())
loader = torch.utils.data.DataLoader(train, batch_size=64, shuffle=True)
net = nn.Sequential(nn.Flatten(), nn.Linear(28*28, 64), nn.ReLU(), nn.Linear(64, 10))
opt = torch.optim.Adam(net.parameters(), lr=1e-3)
loss = nn.CrossEntropyLoss()
for epoch in range(3):
for x, y in loader:
opt.zero_grad()
loss(net(x), y).backward()
opt.step()
print(f"epoch {epoch} → ok") # you'll watch the loss number dropFour evolution stages — pick where to start.
Each plan is a stage in a small network growing up. Start at one neuron; upgrade as your curiosity adds layers.
Starter Access
- Full video course access
- Basic AI & neural-network explanations
- Perceptron introduction
- Lifetime access
Standard Course
- Everything in Starter
- Google Colab exercises
- Simple image & text models
- Homework assignments
- Completion certificate
Pro Learning Path
- Everything in Standard
- Additional practice projects
- Build your first AI model project
- Assignment review
- Basic Q&A support
Premium Mentorship
- Everything in Pro
- 1:1 online session
- Project reviews
- Personalized AI learning plan
- AI career guidance
layer 11 · supporting signals
Learning resources.
-
resource · 01
AI in plain English
A 12-min companion read written for total beginners.
-
resource · 02
Glossary of 40 terms
Words you'll meet — defined once, in a sentence each.
-
resource · 03
Cheat-sheet PDF
One page. The whole training loop, visually.
-
resource · 04
Beginner notebook pack
Five Colab notebooks you can open immediately.
Latest AI insights.
-
fundamentals
Why your first model should be embarrassingly small
Big networks hide the lessons. Start tiny and you'll understand the gradient.
→ read note -
vision
Pixels are just numbers in a costume
How a black-and-white digit becomes a vector your model can read.
→ read note -
career
What 'know AI' actually means in 2026
Hiring signals from real listings — what beginners are expected to demonstrate.
→ read note
Frequently asked questions.
-
No. The first week is about intuition. Code shows up by week two, and only in small, friendly notebooks.
-
Multiplication, addition, and a curve that squashes numbers. That's enough to begin. Anything heavier appears optionally and visually.
-
Fair. We're upfront: we're a young studio, not a global academy. We make that obvious so your expectations match reality. Lifetime access protects you if we keep improving.
-
Yes, from Standard upward. It says you finished the work — not that you're suddenly an engineer. We don't oversell it.
-
14-day refund window on any plan, no questions, as long as you actually try the first module.
A small studio, three weeks in.
SkillBridge Training is a young online learning studio. We launched three weeks ago with a single bet: most "intro to AI" material is written for people who already half-understand it.
We're building the course we wish we'd had — calm, visual, and honest about what's hard. We're not an academy. We're not the future of education. We're a small team, careful with your time, iterating in public.
If you finish a module and something still feels foggy, that's a bug in our explanation, not in you. Write us. We'll fix it in the next pass.
- age3 weeks
- teamsmall, hands-on
- focusAI for beginners
- stylevisual + practical
- promiseno overclaims
The final node — say hello.
-
phone+1 (727) 683-8000
-
emailcontact@skillbridge-training.com
-
address1100 E Sheldon St
Prescott, AZ 86301