🧮 Interactive Labs — edit code, see live output

Variables — assignment and reassignment

x = 5 binds the name x to the int 5. Reassign x = 'hello' rebinds it to a string. Python is dynamically typed.

python

Loops — for, while, comprehensions

for iterates over any iterable. while runs until a condition is false. Comprehensions build lists in one line.

python

Functions, args, kwargs, defaults

def name(params): ... return value. *args collects positional, **kwargs collects keyword.

python

Lists — append, slice, sort, comprehensions

Lists are ordered, mutable, indexed. The most-used Python data structure.

python

Dictionaries — key/value storage

dicts map keys to values. O(1) lookup. Insertion-ordered since Python 3.7.

python

if / elif / else + ternary

Conditional execution. Truthy values run the if branch; falsy values run else.

python

Classes — __init__, self, methods

Blueprints for objects. __init__ runs on creation. self is the instance. Methods are functions in the class.

python

try / except / finally

Errors are raised, caught, handled. Don't swallow exceptions silently — at least log them.

python

Strings — f-strings, methods, slicing

Strings are immutable sequences of characters. f-strings are the modern way to format them.

python

Comprehensions — list, dict, set, generator

One-line loops that build collections. Faster and more Pythonic than for+append.

python

Common tags — h1, p, ul, a, img, div

The 6 tags you'll use 90% of the time. Change the tag name, see the rendering change.

html

Forms — input, label, button

Forms collect user input. Each input needs a name and a label.

html

Tables — the right way (not for layout!)

Use tables for TABULAR DATA only (not for page layout). thead, tbody, tr, th, td.

html

Semantic tags — header, nav, main, article, footer

Semantic tags describe the CONTENT, not just the look. Better for SEO and accessibility.

html

Links and images — the web's basics

Links connect pages. Images display pictures. Both use URL attributes (href / src).

html

HTML entities — < > & © —

Special characters that need a code because they'd otherwise be parsed as HTML.

html

HTML5 boilerplate — the complete starting template

Every HTML page starts with this. DOCTYPE, html, head with meta tags, body. Copy-paste this as your starting point.

html

Complete dashboard page — sidebar + KPIs + charts + table

The full HTML structure of a data-science dashboard. Semantic tags: header, aside, main, section, article.

html

Complete form — all input types + validation

Every input type in one form: text, email, password, number, range, date, checkbox, radio, select, textarea, file.

html

Complete data table — caption, thead, tbody, colspan

A proper data table with caption, header, body, footer, and cells that span multiple columns.

html

Article/blog page — complete content layout

A blog post or article page with header, metadata, content blocks, code blocks, and a footer.

html

Login page — complete auth form

A complete login page with email/password, remember me, forgot password, and social login buttons.

html

color — change the text color

The most basic CSS property. Change the color value and watch the text recolor instantly.

css

background — fill the element's box

background-color fills the box. background-image uses an image. background gradient blends colors.

css

border — outline the element

border needs 3 values: width, style, color. Change any of the three.

css

box-shadow — give the card depth

The example from your message. 5 values: x-offset, y-offset, blur, spread, color.

css

margin vs padding — the box model

margin is OUTSIDE the border (space between elements). padding is INSIDE the border (space between border and content).

css

font-family, font-size, font-weight — typography

Change the typeface, size, and boldness. Use system fonts for instant loading.

css

display — block, inline, none

Controls how an element fits in the page flow. block = full width. inline = only as wide as content. none = hidden.

css

position — static, relative, absolute, fixed

Controls how an element is positioned. The most confusing CSS property — play with it.

css

flexbox — the modern layout system

display: flex turns a parent into a flex container. justify-content and align-items control alignment.

css

grid — 2D layouts

display: grid creates a 2D grid. grid-template-columns defines columns. The most powerful CSS layout.

css

transition — smooth animations

transition: <property> <duration> <easing>. Makes changes animate instead of snapping.

css

z-index — stacking order

z-index controls which element appears on top when they overlap. Higher = on top. Only works on positioned elements.

css

border-radius — round the corners

Rounds each corner. One value = all corners. Four values = top-left, top-right, bottom-right, bottom-left.

css

:hover, :focus, :active, :nth-child

Pseudo-classes target elements based on state or position. Hover, focus, nth-child are the most common.

css

@media — responsive design

@media (max-width: 600px) applies styles only on small screens. Resize your browser to see.

css

Cascade + specificity — which rule wins?

When multiple rules target the same element, the browser picks the winner by specificity. IDs beat classes, classes beat tags.

css

CSS Reset — start every project with this

Remove default browser margins, set box-sizing: border-box on everything. The FIRST 4 lines of every CSS file.

css

CSS Variables (custom properties) — :root

Define colors/sizes once in :root, use var(--name) everywhere. Change once, updates everywhere.

css

Full page layout — header + sidebar + main + footer

The classic dashboard layout: fixed header on top, sidebar on left, main content fills the rest, footer at bottom.

css

Navbar — horizontal navigation bar

Logo on the left, links on the right. Sticky at the top. The most common UI element.

css

Responsive card grid — auto-fit

Grid that automatically shows 1/2/3/4 columns based on screen width. No media queries needed!

css

Form styling — inputs, labels, focus, validation

Make forms look professional: full-width inputs, clear labels, focus rings, error states.

css

Data table — striped rows, hover, sortable header

The table style every dashboard needs: striped rows, hover highlight, bold header, right-aligned numbers.

css

Alert boxes — success, error, warning, info

4 colored boxes for user feedback. Each has an icon, a title, and a message.

css

Loading spinner — pure CSS animation

A spinning circle that shows 'loading in progress'. No JavaScript, no images — pure CSS.

css

Chart container — figure + caption for plots

A styled wrapper for matplotlib/plotly charts. White card, title, the chart, and a caption.

css

Modal dialog — overlay + centered box

A dialog that floats above the page. Dark overlay behind, centered white box, close button.

css

Badges + tags — small status labels

Pills, dots, and tags for status indicators. 'Trained', 'v2.1', 'GPU', 'beta' — small colored labels.

css

Footer — bottom of the page

Multi-column footer with links, social icons, and copyright. The pattern every website uses.

css

Tabs — switch between panels

Tab navigation that switches content panels. Pure CSS using :target or radio inputs (no JS needed for basic tabs).

css

Full dashboard layout — sidebar + KPIs + charts + table

The complete data-science dashboard: sidebar nav, KPI row, 2-column charts, data table.

css

Responsive form page — centered card, mobile-friendly

A complete login/register page: centered card, responsive, branded. The pattern every auth page uses.

css

Your first route — @app.get('/path')

A route is a function FastAPI calls when a matching HTTP request arrives. Decorator specifies method + path.

fastapi

Path parameters — {id} becomes a function arg

Wrap a segment of the path in {braces} and FastAPI extracts it. Type hints validate.

fastapi

Query params — ?key=value after the path

Function args that AREN'T path params become query params. Optional with defaults.

fastapi

Pydantic request bodies — automatic validation

Define a Pydantic model with typed fields. FastAPI validates the JSON body, returns 422 if invalid.

fastapi

Jinja2 templates — render HTML on the server

TemplateResponse renders an HTML file with variables. base.html + child.html pattern.

fastapi

Dependency injection — Depends()

Share setup/teardown across routes. get_db is the classic example — opens a DB session per request.

fastapi

Forms — Form() instead of JSON body

HTML forms send application/x-www-form-urlencoded, not JSON. Use Form(...) to receive.

fastapi

Static files — CSS, JS, images

Mount a directory as a static file server. Files are served as-is, no processing.

fastapi

WebSocket — real-time bidirectional

Persistent connection. Server can push to client without the client asking. Used for live updates, chat, notifications.

fastapi

Middleware — runs before/after every request

Wrap the entire app. Each request passes through before reaching the route, response passes through after.

fastapi

Load a .pkl model file (joblib + pickle)

The first step of any benchmarking engine: load the model the user uploaded. joblib.load() and pickle.loads() both work.

python

Run predict() on a loaded model

Once the model is loaded, call .predict(X) to get predictions. This is the core of inference benchmarking.

python

Measure training time (time.perf_counter)

The 'benchmark' part: how long does model.fit() take? Use time.perf_counter() for high-precision timing.

python

Measure inference time + throughput

How fast can the model make predictions? Measure inference time, per-sample latency, and throughput (samples/sec).

python

Regression metrics — MAE, MSE, RMSE, R²

Compute the 4 standard regression metrics. Each tells you something different about model quality.

python

Classification metrics — accuracy, precision, recall, F1

For classification models: accuracy, precision, recall, F1, confusion matrix. Each catches a different kind of failure.

python

Cross-validation — more honest than a single split

Single train/test split = lucky or unlucky. 5-fold CV runs 5 different splits and averages — more honest.

python

Handle ANY model type (sklearn, xgboost, custom)

A benchmarking engine must work with ANY .pkl — not just LinearRegression. Use duck typing: check for .predict().

python

FastAPI endpoint — accept .pkl file upload

The FastAPI route that receives the .pkl file + form fields. Uses UploadFile = File() + Form().

fastapi

HTML form — upload .pkl + select dataset

The form users see. File input + dataset dropdown + framework dropdown. enctype=multipart/form-data is CRITICAL.

html

Full benchmark engine — putting it all together

The complete benchmarking function: load .pkl → detect model type → load dataset → predict → time → metrics → return JSON. This is what OpenBenchML does.

python

Save benchmark results + compare models

Run benchmarks on multiple models, save results to a DataFrame, compare side-by-side. This is how leaderboards work.

python

Error handling — when the .pkl is broken or wrong

Real users upload broken files. Your benchmark engine must handle: corrupt .pkl, wrong model type, feature mismatch, predict() crashes.

python

Preprocessing mindset — the 7 questions to ask ANY dataset

Before touching ML, ask these 7 questions of ANY dataset. This is how senior data scientists think — not 'load + fit'.

python

Imputation — when to use mean vs median vs mode vs KNN vs MICE

Missing values are NOT a 'fill with mean' problem. Different imputation strategies for different situations. Know when to use each.

python

Encoding — OneHot vs Label vs Target vs Frequency vs Embedding

Categorical encoding is NOT just pd.get_dummies(). 5 strategies, each for a different situation. Pick wrong = garbage model.

python

EDA mindset — 8 questions that reveal what the data is hiding

Exploratory Data Analysis is not 'df.describe()'. It's a systematic investigation. 8 questions that uncover patterns, leaks, and ML-readiness.

python

EDA visualization — 6 plots that tell you everything

Not 'make a chart'. 6 specific plots, each answering a different question. This is how you SEE the data before modeling.

python

Linear Regression maths — the equation behind .fit()

model.fit() is not magic. It solves y = Xβ + ε using least squares. Understand the equation, the cost function, and the geometry.

python

Gradient Descent — the engine behind neural networks

Most ML models (neural nets, logistic regression, SVMs) use gradient descent, not closed-form solutions. Understand the algorithm.

python

Probability for ML — Bayes, distributions, maximum likelihood

ML is applied probability. Understand Bayes' theorem, common distributions, and maximum likelihood estimation.

python

Model selection — which algorithm for which problem?

Not 'try RandomForest'. A decision framework: data size, feature type, interpretability need, linearity assumption. Pick the RIGHT model.

python

Overfitting — the #1 reason ML models fail in production

Understand the bias-variance tradeoff, why train > test means overfitting, and 5 techniques to fix it. This is what separates seniors from freshers.

python

Full ML workflow — the senior data scientist's process

Not 'load + fit + predict'. The REAL workflow: 10 steps from raw data to production-ready model. This is what senior DS actually do.

python

Logistic Regression visualized — sigmoid + decision boundary (iris)

SEE the sigmoid curve, SEE the decision boundary move as you change C. Uses the iris dataset. The maths becomes obvious when you can visualize it.

python

PCA visualized — variance + projection (iris + digits)

SEE PCA reduce 4D iris to 2D. SEE the scree plot. SEE 64D digits collapse to 2D. The geometry of dimensionality reduction.

python

SVM visualized — hyperplane + margins + support vectors (iris)

SEE the SVM hyperplane, the margin, and the support vectors. Compare linear vs RBF kernel. See how C controls the margin width.

python

Decision Tree visualized — the splits + overfitting (iris)

SEE the actual tree structure with its split rules. SEE how depth 3 vs depth 10 changes the decision regions. The most interpretable model.

python

Random Forest visualized — many trees + feature importance (wine)

SEE how 100 trees vote. SEE feature importance. SEE OOB error stabilize. The wine dataset (13 features, 3 classes).

python

KNN visualized — the K nearest neighbors + decision boundary (iris)

SEE how K=1 creates jagged boundaries (overfit) and K=15 creates smooth boundaries (underfit). The simplest ML algorithm, visualized.

python

K-Means visualized — centroid movement + elbow plot (iris)

SEE the centroids move iteration by iteration. SEE the elbow plot to pick K. Unsupervised learning made visual.

python

Gradient Boosting visualized — sequential learning + residuals (breast cancer)

SEE how each tree corrects the PREVIOUS tree's errors. SEE how learning rate + n_estimators interact. The competition-winning algorithm.

python

Interactive: Linear Regression — drag the slope + intercept

DRAG the slope and intercept sliders. SEE the line move in real-time. SEE the MSE (cost) change. This is how you build intuition for what .fit() does.

interactive

Interactive: Logistic Regression — drag the sigmoid + decision boundary

DRAG the weight + bias sliders. SEE the sigmoid curve shift. SEE the decision boundary move. SEE how probability changes for each data point.

interactive

Interactive: KNN — drag K and watch the boundary change

DRAG the K slider. SEE the decision boundary morph from jagged (K=1) to smooth (K=20). SEE how each K affects the classification of a new point.

interactive

Interactive: Overfitting — drag polynomial degree + see train vs test

DRAG the polynomial degree slider. SEE the curve go from straight line (underfit) to wiggly mess (overfit). SEE train vs test error diverge.

interactive

Interactive: K-Means — drag K and watch centroids + clusters

DRAG the K slider. SEE centroids move. SEE clusters form. SEE the elbow plot update. Find the natural number of clusters.

interactive

Interactive: Bias-Variance Tradeoff — the fundamental ML curve

DRAG the model complexity slider. SEE bias decrease, variance increase, and total error form a U-shape. The sweet spot is the minimum.

interactive