How To Become A Machine Learning

Engineer
A Comprehensive Roadmap With Courses

Photo by Jaromír Kavan on Unsplash

https://towardsdatascience.com/how-to-become-a-machine-learning-engineer-
e420e134c0a3
Kurtis Pykes
Jul 19, 2021

A large portion of feeling more fulfilled in your career is from the

feeling of progress. For a while, that feeling has been void for
me. Yes, I’ve still been landing freelance machine learning
contracts but not exactly the ones I want — with all due respect
to my current clients. While I’m grateful for the opportunities
I’ve been receiving, I know that I must improve if I want to reach
the goals I’ve set for my career.
I didn’t become a Freelancer to do work I
don’t want to do.

Consequently, I conducted some research on what skills are

most necessary to become a Machine Learning Engineer. After, I
looked up the most effective courses (from experience and
popular opinion) to upskill in each area. The idea was to gain an
insight into what skills I’m currently lacking as a Machine
Learning Engineer so I can put my focus into improving in those
area’s which in turn will make me a better ML engineer and in
return, improve my odds of landing more freelance projects I
want.

Note: In Essential Skills for Machine Learning Engineers,

I covered each of these concepts hence this article will focus
more on the exact courses to take to upskill in each area.

Computer Science

Working software is the outcome of a successful end-to-end

Machine Learning project. Thus, ML engineers are expected to
have good knowledge of basic computer science fundamentals
since they require excellent software engineering skills to create
working software. The courses to build up your computer
science fundamentals were suggested by Dhav Patel
from codebasics YouTube channel.

 Digital Information
  The Internet

 Programming

 Algorithms

Programming Language

The next and most obvious step is to learn a programming

language. If the output of a Machine Learning engineer is
deliverable software then you’ve got to learn how to create
software. This requires knowledge of a programming language.

Python is the most popular language for Machine Learning. I’ve

already created a list of the Best resources to learn Python
for Machine Learning and Data Science so be sure to
check that out. Depending on where you work, some companies
may expect you to have knowledge of other languages such as
Java and C++ (mainly because they are faster than Python). I
personally like to use Codeacademy to learn programming
languages. Here are the respective courses for C++ & Java:

 Learn C++

 Learn Java

Note: Learn one language first (probably Python) then

move on.

Data Structures & Algorithms

Data Structures & Algorithms (DSA) are often ignored when we
talk about Machine Learning, but this isn’t a true reflection of its
importance. DSA covers solutions to standard problems in detail
and provides us with a better understanding of how efficient it is
to use each one.

Also, it teaches us the science behind evaluating the efficiency of

an algorithm which permits us to decide the best solution to our
problem from a variety of choices. This is extremely important
for a Machine Learning engineer because sometimes we may be
required to write our own algorithms, hence why a good
foundation in DSA is essential.

The best course and places to practice Data Structures &

Algorithms (by popular demand) include:

 Algorithms and Data Structures in Python (Udemy)

 HackerRank Algorithms (Problem Solving)

 HackerRank Data Structures (Problem Solving)

 Leetcode Algorithms (Problem Solving)

Note: The langauge you use to learn Algorithms and Data

Structures isn’t important. Be more keen to understand the
fundamental principles.

Relational Databases
Data is a prerequisite for Machine Learning; No data, no
Machine Learning. Although the field is branching out to other
areas that involve unstructured data (text, images, video, etc),
it’s still safe to say that most of the data used for Machine
Learning is structured. Structured data typically lives in a
relational database and all relational databases use SQL. In fact,
the majority of big data tools use SQL so it’s worth learning.

 Learn SQL Basics for Data Science (Coursera)

 SQL for Data Science (Coursera)

 The Complete SQL Bootcamp 2021: Go From Zero to

Hero (Udemy)

 The Ultimate MySQL Bootcamp: Go From SQL

Beginner to Expert (Udemy)

 Learn SQL (Codeacademy)

Note: Select and complete 1 then move on!

Mathematics & Statistics

Machine Learning involves a lot of math. Math is what allows

the algorithms we use to unearth the patterns in data so they can
make decisions. Although we would occasionally have to revisit
various math concepts to understand different technologies,
systems, and architectures in our ML career, it’s important to
have a firm foundation, in the beginning, to get us started. The
foundational math can be broken down into the following
categories;
Linear Algebra Courses:

 Mathematics For Machine Learning: Linear Algebra

 Khan Academy: Linear Algebra

Calculus Courses:

 Mathematics For Machine Learning: Multivariate

Calculus

 Khan Academy: Calculus

Probability and Statistics Courses:

 Khan Academy: Statistics and Probability

Note: We do not need to become advanced mathetaticians

because that could take a lifetime. The goal is to have a
good enough foundation to understand various concepts in
Machine Learning.

De Facto Data Science Libraries

As you up the ante on your goal to becoming a Machine

Learning Engineer, there comes a time when you must focus on
the de facto Data Science frameworks because you’re going to be
using them almost every day. For now, we will leave out the
machine learning & deep learning frameworks.

 NumPy
  Pandas

 SciPy

 Matplotib / Seaborn / Plotly (Select one)

Note: Try to use these frameworks to build something.

Machine Learning Algorithms

It wouldn’t make sense to be a Machine Learning Engineer

without knowing machine learning. Like most topics in Machine
Learning, there are a number of courses to learn the actual
Machine Learning algorithms, but for me, the best one
is Machine Learning by Standford University.

Key algorithms to know are:

 Linear Regression

 Logistic Regression

 Decision Tree

 Random Forest

 Naive Bayes

 Support Vector Machines

 K-Nearest Neighbors

 K-Means Clustering

 Dimensionality Reductions Algorithms (i.e. PCA)

  Gradient Boosting Machines (i.e. GBM, XGBoost,
LightGBM, etc)

Note: You also want to learn the Python De Facto

framework for Machine Learning, Scikit-Learn. It’s a
great idea to build something using the framework.

Deep Learning

Deep Learning is a growing subfield of Machine Learning. The

architectures in Deep learning are inspired by the structure and
function of the brain, hence the name “Neural Networks”. A
good foundation in Machine Learning, especially Linear
regression, makes the progression into deep learning a lot more
simple. I’d recommend taking the Deep Learning Specialization
by DeepLearning.ai and learning TensorFlow or PyTorch.

Note: It doesn’t matter which one you pick; choose one

and get really good. To those that decide to learn
TensorFlow, I’d also suggest you consider taking the
TensorFlow Developer certificate — I haven’t done it yet,
but it’s definitely on the cards.

MLOps

MLOps is the latest craze on the Machine Learning block. It’s

the DevOps equivalent for Machine Learning and ML Engineers
should know it. It may motivate you to know that the majority I
am approached by nowadays all have something to do with
MLOps which is a very interesting trend. Some resources to
check out if you would like to learn MLOps include:

 Machine Learning Engineering for Production

(MLOps) Specialization (Coursera)

 Introducing MLOps (Book)

 MLOps (Machine Learning Operations)

Fundamentals (Coursera) — This course is part of the
preparing for Google Cloud certification: Machine
Learning Engineer

Additional Learning

Once you’ve learned the aforementioned skills, you will be ready

to get into work as a Machine Learning engineer. The following
skills are “good to have” and will help you stand out from the
competition so there is some merit in learning each one and
when they’re applicable.

 PySpark

 Hadoop

 Docker

 CI-CD for Machine Learning

 Version Control With Git

 FastApi, Tensorflow Serving

 NoSQL Databases
Final Thoughts

Becoming a Machine Learning Engineer is a tough sport. You’ll

have to commit yourself to develop your skillset so that you can
confidently build and deploy machine learning systems.

There is no point in taking every single course in this article. A

better solution is to find areas where you’re lacking competency
and devote yourself to building up that area. Throughout the
journey, also try to use the resources you’re learning to build
your portfolio either through blogs, vlogs, projects, etc.

Thanks for Reading!

If you enjoyed this article, connect with me by subscribing to

my FREE weekly newsletter. Never miss a post I make about
Artificial Intelligence, Data Science, and Freelancing.