Day, job title Data Scientist, Machine Learning Engineerand AI Engineer It's everywhere – if you're like me, it can be hard to understand what each of them actually do if you're not working in the field.
And there is a title that sounds even more confusing – Quantum Blockchain LLM Robot Engineer (Okay, I made it, but you get points).
There are many roles in the job market that overlap with buzzwords, making it difficult to know where to start if you're interested in a career in machine learning.
This article categorizes the roles of machine learning at the top and explains what each entails. Furthermore, we will explain what you need to prepare them.
Data Scientist
What is it?
Data scientists are the best known role, but have the greatest scope of responsibility.
Generally, there are two types of data scientists.
- It focuses on analysis and experiments.
- Focus on machine learning and modeling.
The former includes implementing A/B tests, conducting deep dives to determine where the business can improve, and suggesting improvements to machine learning models by identifying blind spots. Much of this task is called explanatory data analysis or EDA.
The latter is primarily concerned with the construction of POC machine learning models and decision-making systems that benefit the business. We then work with software and machine learning engineers to deploy these models and monitor their performance.
Many machine learning algorithms usually come from a simpler aspect, resulting in periodic, unsupervised learning models such as:
- xgboost
- Linear and logistic regression
- Random Forest
- k-means clustering
I was a data scientist at an older company, but I mainly created machine learning models and did not perform many A/B tests or experiments. It was a job carried out by data and product analysts.
However, in my current company, data scientists do not build machine learning models, but primarily do deep analysis and measure experiments. Model development is primarily carried out by machine learning engineers.
It all really comes down to the company. Therefore, it is really important to read the job description and make sure it is the right job for you.
What do they use?
As a data scientist, these are generally things you need to know (it's not exhaustive and depends on your role):
- Python and SQL
- git and github
- Command Line (Bash and ZSH)
- Statistics and Mathematics Knowledge
- Basic machine learning skills
- A little cloud system (AWS, Azure, GCP)
If this role is interested in you, there is a roadmap to become a data scientist that you can see below.
Machine Learning Engineer
What is it?
As the title suggests, machine learning engineers are about building machine learning models and deploying them into production systems.
Originally from software engineering, it is now a unique job/title.
The key difference between machine learning engineers and data scientists is that machine learning engineers deploy algorithms.
Chip Huyen, a leading AI/ML practitioner, puts it:
That's the goal of data science Generate business insightsML engineering's goal is Turn your data into products.
Data scientists often come from strong mathematics, statistics, or economics backgrounds, and machine learning engineers can see that they are more from science and engineering backgrounds.
However, there is a significant overlap in this role, allowing some companies to bundle data scientists and machine learning engineer positions into a single job.
The work of machine learning engineers is usually seen in more established tech companies. However, it slowly grows popular over time.
Additionally, there is more expertise within the roles of machine learning engineers:
- ML Platform Engineer
- ML Hardware Engineer
- ML Solutions Architect
If you're a beginner, don't worry about these. Because they are quite niche and are only relevant after a few years of experience in this field. I wanted to add these, so I know a variety of options.
What do they use?
The technology stack is very similar to machine learning engineers as well as data scientists, but with more software engineering elements.
- However, Python and SQL may require other languages by some companies. For example, in my current role, I need rust.
- git and github
- Bash and ZSH
- AWS, Azure or GCP
- Basics of software engineering such as CI/CD, MLOPS, Docker, and more.
- Excellent machine learning knowledge, ideally local expertise.
AI Engineer
What is it?
This is a new title born out of all the AI hype ongoing, and to be honest, it's a weird title and I don't think it really needs it. Machine learning engineers often play the role of AI engineers in most companies.
Most AI engineers' roles are actually about Genai, not AI as a whole. This distinction usually makes no sense for people outside of the industry.
However, AI covers almost all decision-making algorithms and is larger than the machine learning field.
The current definition of AI engineers is those who support their business primarily using LLM and Genai tools.
Unless you are primarily in the lab, you will not develop the underlying algorithm from scratch. Many of the top models are open source, so there's no need to reinvent the wheel.
Instead, they will first focus on adapting and building the product, then worry about fine-tuning the model. So they're woo
It's much closer to traditional software engineering than the current role of machine learning engineers. Many machine learning engineers run as AI engineers, but their jobs are new and have not yet been fully embodied.
What do they use?
This role has evolved considerably, but generally requires a good knowledge of all the latest trends in Genai and LLM.
- Solid software engineering skills
- Python, SQL, BackEnd Langauges like Java and Go are useful
- CI/CD
- git
- LLMS and Transformers
- Rags
- Fast engineering
- Basic model
- Fine adjustments
We also recommend checking out DataCamp's Associates AI Engineer for Data Scientist Track. This sets you up well for a career as a data scientist. This is linked to the description below.
Research Scientist/Engineer
What is it?
Previous roles were primarily industry positions, but these two are based on research.
The role of the industry is primarily associated with the business, and is all about creating business value. Whether using linear regression or trans models, the impact is what matters, and not necessarily the method.
The purpose of the research is to broaden current knowledge capabilities, both theoretical and practical. This approach revolves around scientific methods and deep experiments in niche fields.
The differences between research and industry are ambiguous and often overlapping. For example, many top research labs are actually large companies.
- Meta-study
- Google AI
- Microsoft AI
These companies initially began solving business problems, but now they have dedicated research departments, allowing them to tackle industry and research issues. One begins, the other end is not always clear.
If you are interested in exploring more depth in the differences between research and industry, we recommend reading this document. This is the first lecture on Stanford's CS 329S, Lecture 1: Understanding Machine Learning Production.
Generally, there are more industry jobs than research, as only large companies can pay data and computing costs.
Anyway, as a research engineer or scientist, you essentially work on cutting-edge research and push the boundaries of machine learning knowledge.
There is a slight distinction between the two jobs. As a research scientist, you need a doctorate, but this does not necessarily apply to a research engineer.
Research engineers usually implement the theoretical details and ideas of research scientists. This role is usually lies across established research companies. In most situations, research engineers and scientists work the same.
Companies may offer research scientist titles as they give them more “influence” and are more likely to get a job.
What do they use?
This is similar to machine learning engineering, but the depth of knowledge and qualifications is often increased.
- Python and SQL
- git and github
- Bash and ZSH
- AWS, Azure or GCP
- Basics of software engineering such as CI/CD, MLOPS, Docker, and more.
- Excellent machine learning knowledge and expertise in cutting-edge fields such as computer vision, reinforcement learning, and LLM.
- PhD or at least a Master's degree in related discipline.
- Research experience.
This article just scratched the surface of machine learning roles, and these four or five I mentioned have more niche work and expertise.
I always recommend starting your career by putting your feet in the door and turning in the direction you want to go. This strategy is much more effective than tunnel vision for only one role.
One more!
We offer 1:1 coaching calls that let you chat about what you need, whether you're just figuring out what you need, whether you're grasping the next step or not. I'm here to help you move forward!
1:1 Mentoring call with Egor Howell
Career guidance, job advice, project help, resume reviewTopmate.io
