Train, deploy and maintain machine learning models that work reliably in real products.
You have reached the end of the beginner stages. Prove what you learned with a project.
You have reached the end of the intermediate stages. Prove what you learned with a project.
You have reached the end of the advanced stages. Prove what you learned with a project.
Train and compare two regression models on an open dataset and explain which is better and why.
Build a text classifier and report precision and recall, not just accuracy.
Fine-tune a pretrained vision model on a small set of your own photos.
Engineer features from raw activity data and show which ones matter most.
Serve a trained model in a container with an API, tests and tracked experiments.
Simulate changing input data and alert when live predictions stop matching training data.
The same roadmap as a reading list: every stage, what each topic covers, and which ones are optional.
Write clean Python and work with arrays, the building block of every ML library.
Vectors, matrices and how data and model weights are represented and combined.
Derivatives and gradients: how a model learns by nudging its weights.
Distributions, uncertainty and the reasoning behind evaluation.
Regression and classification: learn from labelled examples to predict new ones.
The standard library for classic models, pipelines and preprocessing.
Train/test splits, cross-validation, the right metric for the problem, and overfitting.
Clustering and dimensionality reduction for data without labels.
Layers, activations, loss functions and backpropagation, understood visually.
The most widely used deep learning framework in research and industry.
A mature alternative with strong tooling for mobile and the browser.
Pretrained models for text and images, and how to adapt them to your task.
Load, clean, split and check data so the model learns the right thing.
Create inputs that make patterns easier for a model to learn.
Pull training data straight from the databases where it lives.
Know exactly which data and code produced each model.
Record parameters, metrics and artefacts so results can be compared and repeated.
Put a model behind an API with input validation and sensible latency.
Package a model and its dependencies so it runs the same everywhere.
Watch live predictions and data, and retrain when the world changes.
Frame the problem, pick the data, model, serving and feedback loop as one system.
Managed training, tuning and deployment on a cloud provider.
Adapt pretrained models efficiently instead of training from scratch.
Measure fairness, protect privacy and explain what a model can and cannot do.
They turn models into dependable software: preparing data, training and evaluating models, deploying them and keeping them accurate once real users rely on them.
Data scientists focus more on analysis, experiments and insight. ML engineers focus more on building, deploying and scaling models. Many teams blend the two.
Enough linear algebra, calculus and statistics to understand how models learn and how to judge them. You do not need to derive everything by hand.
© 2026 KamerLink. All rights reserved.
