Skip to content
T_
TENSORCRAFT
Themes
Pricing
Start now →
// bridge system
useState()
→
Model Weights
·
Event Propagation
→
Forward Pass
·
Array.map()
→
Tensor Operation
·
React diff
→
Loss Function L = Σ(y−ŷ)²
·
transition-duration
→
Learning Rate η
·
CSS clamp()
→
σ(x) Activation
·
Re-render cycle
→
Training Epoch
·
Event bubbling
→
Backpropagation ∂L/∂w
·
useCallback
→
Gradient Caching
·
Promise.all()
→
Batch Inference
·
Redux store
→
Weight Matrix
·
DevTools profiler
→
Loss Landscape
·
useState()
→
Model Weights
·
Event Propagation
→
Forward Pass
·
Array.map()
→
Tensor Operation
·
React diff
→
Loss Function L = Σ(y−ŷ)²
·
transition-duration
→
Learning Rate η
·
CSS clamp()
→
σ(x) Activation
·
Re-render cycle
→
Training Epoch
·
Event bubbling
→
Backpropagation ∂L/∂w
·
useCallback
→
Gradient Caching
·
Promise.all()
→
Batch Inference
·
Redux store
→
Weight Matrix
·
DevTools profiler
→
Loss Landscape
·
//
Report Issue
Report Issue
Category
Bug
Feature Request
Content Issue
Payment Problem
Other
Description
0/2000
Severity
low
medium
high
critical
Screenshot
+
Attach screenshot (optional)
Submit Report
page url + browser info captured automatically
// Neon Protocol — 11 MODULES
MODULE MAP
Track your journey from frontend engineer to ML engineer.
0
Completed
1
Active Now
—
Avg Score
>
M01
ML Through Vision
ACTIVE
Noor joins the resistance and learns the fundamentals of computer vision through surveillance analysis. Images are just pixel arrays — and she's been working with them all along.
>
Seeing Like a Machine
2
Pixels & Tensors
3
Color Channels
4
Image Operations
5
Your First Classifier
○
M02
Image Data
Descend into Neon City's underground labs to master image dataset construction. You'll learn to collect, augment, normalize, balance, and efficiently load image data — transforming raw surveillance footage into training-ready datasets for the Protocol Sentinel.
○
M03
Convolutional Neural Networks
Deep in the resistance bunker, Noor and Zara build the Protocol Sentinel's visual cortex. You'll master convolution operations, pooling, CNN architecture, training pipelines, and feature visualization — teaching a neural network to distinguish surveillance camera types from Meridian's grid.
○
M04
Object Detection
Amir reveals how Meridian's surveillance drones track targets in real-time. The resistance reverse-engineers object detection — from bounding boxes to YOLO — and builds something faster that runs entirely in the browser. For the first time, you see the drones before they see you.
○
M05
Segmentation
A drone detector tells you what — segmentation tells you where. Zara teaches Noor to classify every pixel in a street scene, distinguishing individual objects and generating precise masks. The resistance can finally map the full surveillance landscape at pixel-level resolution.
○
M06
Feature Extraction
Amir reveals that Meridian stores features, not images — compact numerical fingerprints that encode what makes each frame unique. Noor learns to extract embeddings, build similarity search, and navigate the geometry of visual meaning.
○
M07
Transfer Learning
The midpoint revelation. Amir downloads Meridian's own model weights from a server room deep in the corporate district. Two million credits of training — repurposed with 500 images and a fine-tuning loop. The resistance turns the enemy's weapons against them.
○
M08
Real-Time Detection
Protocol Sentinel goes live on a test feed. The team fractures over architecture — Noor wants pure browser, Amir wants hybrid. Dilan settles it: 'Build both. Let the data decide.' You build a full video processing pipeline with WebGL acceleration that runs inference at 30fps.
○
M09
Generative Vision
Zara's module. Adversarial patches that confuse Meridian's cameras — art that literally fights back. But Dilan halts deployment for ethical review: 'We deploy this responsibly or we don't deploy it at all.' The patches are cryptographically bound to Meridian's architecture fingerprints only.
○
M10
Deployment
Amir leads deployment. Meridian is quiet — no new drones, no model updates, no probes. The silence is deliberate. The team uses the calm strategically: optimizing, stress-testing, hardening Protocol Sentinel for every resistance device in Neon City.
○
M11
Capstone — Protocol Sentinel
The final mission. Protocol Sentinel deploys across hundreds of devices. Evidence of Meridian's illegal profiling broadcasts across every screen in Neon City. Watch the watchers.
Neon Protocol — Module Map | Tensorcraft