All Roadmaps / Machine Learning Engineer

Machine Learning Interactive Graph Curriculum

Machine Learning Engineer

A master curriculum from mathematical foundations and statistical intuition to production scikit-learn architectures, ensemble methods, and automated MLOps pipelines.

5 Phases

17 Topics

~164h Total Time

Study Progress: 0 of 17 completed • 0h of ~164h 0%

Status:

Level:

🎯

Recommended Next Step Beginner ⏱️ 3 hours

Intelligent Agents & Problem Formulation

Graph Legend:

Recommended

Optional

Beginner

Intermediate

Advanced

Done • In Progress • Use J/K or click cards to explore

Curriculum Start

01 Phase 01 0/3 completed

Phase 1: Mathematical Foundations for ML

Core linear algebra, multivariable calculus, and probabilistic modeling essential for algorithmic intuition.

beginner Recommended • 2 weeks

Linear Algebra & Matrix Transformations

Vector spaces, dot products, matrix multiplication, rank, determinants, eigenvalues, eigenvectors, and Singular Value Decomposition (SVD).

intermediate Recommended • 2 weeks

Multivariable Calculus & Gradient Optimization

Derivatives, partial derivatives, directional gradients, Jacobian and Hessian matrices, and optimization via Gradient Descent variants.

intermediate Recommended • 2 weeks

Probability Distributions & Bayesian Inference

Random variables, probability mass/density functions (Normal, Bernoulli, Poisson), Bayes' Theorem, Maximum Likelihood Estimation (MLE), and Maximum A Posteriori (MAP).

02 Phase 02 0/3 completed

Phase 2: Python Tooling & Scientific Computing

Professional engineering environment, vectorized computation with NumPy, and structured data handling with Pandas.

beginner Recommended • 1 week

Python Environment & AI/ML Setup

Configuring modern Python (3.12+), virtual environments, pip, poetry, VS Code configuration, and interactive Jupyter notebook workflows.

beginner Recommended • 10 days

NumPy & Vectorized Array Operations

N-dimensional array indexing, memory layout (C vs Fortran order), broadcasting semantics, vectorization, and linear algebra operations.

intermediate Recommended • 2 weeks

Pandas for Data Wrangling & Feature Engineering

DataFrames, Series, handling missing values, grouped aggregations, time series manipulation, merge/join paradigms, and categorical encoding.

03 Phase 03 0/4 completed

Phase 3: Supervised Machine Learning

Core predictive algorithms for continuous estimation and discrete pattern classification.

beginner Recommended • 10 days

Linear & Polynomial Regression

Ordinary Least Squares (OLS), cost functions, bias-variance decomposition, and L1 (Lasso) vs L2 (Ridge) vs ElasticNet regularization.

beginner Recommended • 10 days

Logistic Regression & Classification Metrics

Binary and multiclass classification, sigmoid/softmax functions, cross-entropy log-loss, ROC-AUC, precision, recall, and PR curves.

intermediate Recommended • 2 weeks

Decision Trees & Tree Ensembles (Random Forests, XGBoost)

Information gain, Gini impurity, CART algorithm, bagging (Random Forest), and gradient boosting architectures (XGBoost, LightGBM, CatBoost).

intermediate Recommended • 1 week

Support Vector Machines & Kernel Methods

Maximum margin hyperplanes, soft margin formulation (C parameter), dual formulation, and non-linear kernel tricks (RBF, Polynomial).

04 Phase 04 0/3 completed

Phase 4: Unsupervised Learning & Clustering

Finding latent structure, clusters, and low-dimensional manifolds in unlabeled datasets.

intermediate Recommended • 10 days

Clustering Algorithms (K-Means, DBSCAN, Hierarchical)

Centroid-based clustering (K-Means, K-Means++), density-based clustering (DBSCAN), agglomerative hierarchical clustering, and silhouette evaluation.

advanced Recommended • 10 days

Dimensionality Reduction & PCA

Principal Component Analysis (PCA), explained variance ratio, t-SNE, and UMAP for high-dimensional feature compression and latent visualization.

intermediate Optional • 1 week

Anomaly Detection & Density Estimation

Isolation Forests, One-Class SVMs, and Gaussian Mixture Models (GMM) for fraud detection, outlier identification, and industrial telemetry monitoring.

05 Phase 05 0/4 completed

Phase 5: Evaluation, Validation & MLOps

Industrial-grade validation strategies, model deployment, API creation, and production monitoring.

intermediate Recommended • 1 week

Cross-Validation & Preventing Data Leakage

K-Fold, Stratified K-Fold, TimeSeriesSplit, GroupKFold, target leakage prevention, and Scikit-Learn Pipeline architectures.

intermediate Recommended • 5 days

Model Serialization & ONNX Export

Serializing trained estimators with Joblib, Safetensors, and exporting scikit-learn models to the open-standard ONNX format for C++ / cross-platform inference.

advanced Recommended • 10 days

Production Serving with FastAPI & Docker

Building asynchronous REST inference microservices with FastAPI, Pydantic request validation, Docker containerization, and health check endpoints.

advanced Optional • 1 week

Experiment Tracking & Drift Monitoring

MLflow, Weights & Biases, detecting concept drift, data drift (Evidently AI), and continuous model re-training strategies.

Curriculum Breakdown

Phase 1: Mathematical Foundations for ML

0/3 completed

Core linear algebra, multivariable calculus, and probabilistic modeling essential for algorithmic intuition.

Linear Algebra & Matrix Transformations

beginner ⏱️ 2 weeks

Vector spaces, dot products, matrix multiplication, rank, determinants, eigenvalues, eigenvectors, and Singular Value Decomposition (SVD).

Key Outcomes:

✓ Vector projection and geometric intuition of feature spaces

✓ Eigen-decomposition and covariance matrix transformation

✓ SVD formulation for dimensionality reduction

Links: 3Blue1Brown - Essence of Linear Algebra MIT 18.06 Linear Algebra - Gilbert Strang

Multivariable Calculus & Gradient Optimization

intermediate ⏱️ 2 weeks

Derivatives, partial derivatives, directional gradients, Jacobian and Hessian matrices, and optimization via Gradient Descent variants.

Key Outcomes:

✓ Computing gradients with multivariable chain rule

✓ Formulating loss landscapes and convexity

✓ Batch, Mini-batch, and Stochastic Gradient Descent dynamics

Links: Essence of Calculus - 3Blue1Brown Math for Machine Learning Playlist

Probability Distributions & Bayesian Inference

intermediate ⏱️ 2 weeks

Random variables, probability mass/density functions (Normal, Bernoulli, Poisson), Bayes' Theorem, Maximum Likelihood Estimation (MLE), and Maximum A Posteriori (MAP).

Key Outcomes:

✓ Formulating likelihood functions and log-likelihood

✓ Bayesian prior-to-posterior updates

✓ Hypothesis testing, p-values, and confidence intervals

Links: StatQuest with Josh Starmer - Probability & Statistics Fundamental Math for Data Science - Codecademy

Phase 2: Python Tooling & Scientific Computing

0/3 completed

Professional engineering environment, vectorized computation with NumPy, and structured data handling with Pandas.

Python Environment & AI/ML Setup

beginner ⏱️ 1 week

Configuring modern Python (3.12+), virtual environments, pip, poetry, VS Code configuration, and interactive Jupyter notebook workflows.

Key Outcomes:

✓ Isolated dependency management with venv and pip

✓ Interactive kernel debugging in VS Code

✓ Setting up hardware acceleration (Metal / CUDA if available)

Links: HarvardX CS50's Introduction to Programming with Python Python 3 Official Documentation & Downloads

NumPy & Vectorized Array Operations

beginner ⏱️ 10 days

N-dimensional array indexing, memory layout (C vs Fortran order), broadcasting semantics, vectorization, and linear algebra operations.

Key Outcomes:

✓ Eliminating Python loops with vectorized NumPy expressions

✓ Broadcasting rules across unequal matrix shapes

✓ Memory views vs deep copies in matrix slicing

Links: NumPy Illustrated: The Visual Guide Practice Python on HackerRank

Pandas for Data Wrangling & Feature Engineering

intermediate ⏱️ 2 weeks

DataFrames, Series, handling missing values, grouped aggregations, time series manipulation, merge/join paradigms, and categorical encoding.

Key Outcomes:

✓ Exploratory data analysis (EDA) workflows

✓ Imputation, robust outlier handling, and scaling

✓ One-hot encoding vs target encoding

Links: IBM Data Science Professional Certificate Google Data Analytics Professional Certificate

Phase 3: Supervised Machine Learning

0/4 completed

Core predictive algorithms for continuous estimation and discrete pattern classification.

Linear & Polynomial Regression

beginner ⏱️ 10 days

Ordinary Least Squares (OLS), cost functions, bias-variance decomposition, and L1 (Lasso) vs L2 (Ridge) vs ElasticNet regularization.

Key Outcomes:

✓ Deriving normal equation vs gradient descent solution

✓ Regularization penalties to prevent overfitting

✓ Residual analysis and heteroskedasticity checks

Links: HarvardX: Data Science: Machine Learning Scikit-Learn Generalized Linear Models

Logistic Regression & Classification Metrics

beginner ⏱️ 10 days

Binary and multiclass classification, sigmoid/softmax functions, cross-entropy log-loss, ROC-AUC, precision, recall, and PR curves.

Key Outcomes:

✓ Log-odds and odds ratio interpretations

✓ Balancing precision vs recall trade-offs

✓ Handling imbalanced datasets with class weights & SMOTE

Links: Andrew Ng's Machine Learning Specialization (Stanford / Coursera) StatQuest: Logistic Regression Clearly Explained

Decision Trees & Tree Ensembles (Random Forests, XGBoost)

intermediate ⏱️ 2 weeks

Information gain, Gini impurity, CART algorithm, bagging (Random Forest), and gradient boosting architectures (XGBoost, LightGBM, CatBoost).

Key Outcomes:

✓ Tree splitting criteria and pruning strategies

✓ Bagging vs Boosting variance/bias reduction mechanisms

✓ Hyperparameter tuning: learning rate, max depth, subsample

Links: XGBoost Documentation & Tutorials StatQuest: Random Forests & Gradient Boost

Support Vector Machines & Kernel Methods

intermediate ⏱️ 1 week

Maximum margin hyperplanes, soft margin formulation (C parameter), dual formulation, and non-linear kernel tricks (RBF, Polynomial).

Key Outcomes:

✓ Support vector identification and margin boundaries

✓ Kernel mapping to infinite-dimensional spaces

✓ Computational scalability constraints of kernel SVMs

Links: Scikit-Learn SVM Guide

Phase 4: Unsupervised Learning & Clustering

0/3 completed

Finding latent structure, clusters, and low-dimensional manifolds in unlabeled datasets.

Clustering Algorithms (K-Means, DBSCAN, Hierarchical)

intermediate ⏱️ 10 days

Centroid-based clustering (K-Means, K-Means++), density-based clustering (DBSCAN), agglomerative hierarchical clustering, and silhouette evaluation.

Key Outcomes:

✓ Elbow method and Silhouette analysis for optimal K

✓ Handling arbitrary cluster shapes and noise with DBSCAN

✓ Dendrogram interpretation for hierarchical taxonomies

Links: Scikit-Learn Clustering Overview Visualizing K-Means Clustering

Dimensionality Reduction & PCA

advanced ⏱️ 10 days

Principal Component Analysis (PCA), explained variance ratio, t-SNE, and UMAP for high-dimensional feature compression and latent visualization.

Key Outcomes:

✓ Covariance matrix projection onto orthogonal eigenvectors

✓ Balancing dimensionality reduction with information loss

✓ Non-linear manifold visualization with UMAP

Links: PCA Explained Visually (Setosa) How to Use t-SNE Effectively (Distill.pub)

Anomaly Detection & Density Estimation

intermediate Optional ⏱️ 1 week

Isolation Forests, One-Class SVMs, and Gaussian Mixture Models (GMM) for fraud detection, outlier identification, and industrial telemetry monitoring.

Key Outcomes:

✓ Path-length isolation metrics in Isolation Forest

✓ Expectation-Maximization (EM) algorithm for GMMs

✓ Threshold tuning under extreme class imbalance

Links: Scikit-Learn Anomaly Detection

Phase 5: Evaluation, Validation & MLOps

0/4 completed

Industrial-grade validation strategies, model deployment, API creation, and production monitoring.

Cross-Validation & Preventing Data Leakage

intermediate ⏱️ 1 week

K-Fold, Stratified K-Fold, TimeSeriesSplit, GroupKFold, target leakage prevention, and Scikit-Learn Pipeline architectures.

Key Outcomes:

✓ Preventing pre-processing data leakage inside CV splits

✓ Time series temporal ordering constraints

✓ Nested cross-validation for unbiased hyperparameter evaluation

Links: Scikit-Learn Model Evaluation & Cross-Validation

Model Serialization & ONNX Export

intermediate ⏱️ 5 days

Serializing trained estimators with Joblib, Safetensors, and exporting scikit-learn models to the open-standard ONNX format for C++ / cross-platform inference.

Key Outcomes:

✓ Safe serialization patterns vs insecure pickle vulnerabilities

✓ ONNX Runtime acceleration for low-latency batch scoring

Links: ONNX Runtime: Cross-Platform Machine Learning Accelerator

Production Serving with FastAPI & Docker

advanced ⏱️ 10 days

Building asynchronous REST inference microservices with FastAPI, Pydantic request validation, Docker containerization, and health check endpoints.

Key Outcomes:

✓ Designing low-latency REST endpoints for real-time inference

✓ Multi-worker Uvicorn configurations inside Docker containers

✓ Handling batch requests with asynchronous queuing

Links: FastAPI Documentation Google Cloud ML Engineer Learning Path

Experiment Tracking & Drift Monitoring

advanced Optional ⏱️ 1 week

MLflow, Weights & Biases, detecting concept drift, data drift (Evidently AI), and continuous model re-training strategies.

Key Outcomes:

✓ Logging hyperparameters, artifacts, and evaluation metrics

✓ Statistical tests (KS-test, Population Stability Index) for drift

Links: MLflow: An Open Source Platform for the Machine Learning Lifecycle Explore Azure with OpenAI

Beginner 45 min

Topic Title

Status:

What You Need to Know

Topic explanation description goes here.

Code Blueprint

Progress exported & copied to clipboard!

Machine Learning Engineer

Intelligent Agents & Problem Formulation

Phase 1: Mathematical Foundations for ML

Linear Algebra & Matrix Transformations

Multivariable Calculus & Gradient Optimization

Probability Distributions & Bayesian Inference

Phase 2: Python Tooling & Scientific Computing

Python Environment & AI/ML Setup

NumPy & Vectorized Array Operations

Pandas for Data Wrangling & Feature Engineering

Phase 3: Supervised Machine Learning

Linear & Polynomial Regression

Logistic Regression & Classification Metrics

Decision Trees & Tree Ensembles (Random Forests, XGBoost)

Support Vector Machines & Kernel Methods

Phase 4: Unsupervised Learning & Clustering

Clustering Algorithms (K-Means, DBSCAN, Hierarchical)

Dimensionality Reduction & PCA

Anomaly Detection & Density Estimation

Phase 5: Evaluation, Validation & MLOps

Cross-Validation & Preventing Data Leakage

Model Serialization & ONNX Export

Production Serving with FastAPI & Docker

Experiment Tracking & Drift Monitoring

Curriculum Breakdown

Phase 1: Mathematical Foundations for ML

Linear Algebra & Matrix Transformations

Multivariable Calculus & Gradient Optimization

Probability Distributions & Bayesian Inference

Phase 2: Python Tooling & Scientific Computing

Python Environment & AI/ML Setup

NumPy & Vectorized Array Operations

Pandas for Data Wrangling & Feature Engineering

Phase 3: Supervised Machine Learning

Linear & Polynomial Regression

Logistic Regression & Classification Metrics

Decision Trees & Tree Ensembles (Random Forests, XGBoost)

Support Vector Machines & Kernel Methods

Phase 4: Unsupervised Learning & Clustering

Clustering Algorithms (K-Means, DBSCAN, Hierarchical)

Dimensionality Reduction & PCA

Anomaly Detection & Density Estimation

Phase 5: Evaluation, Validation & MLOps

Cross-Validation & Preventing Data Leakage

Model Serialization & ONNX Export

Production Serving with FastAPI & Docker

Experiment Tracking & Drift Monitoring

No matching topics found

Topic Title

What You Need to Know

Code Blueprint

Key Milestones & Checkpoints

Recommended Prerequisites

Personal Study Notes

Video Tutorial

Official Guides & Documentation

Further Reading & Standards

Related EncodeEdge Tutorials

Keyboard Shortcuts

Reset All Progress?