In 2025 If you’re hoping to break into tech or pivot into a new job family, figuring out which career path is right for you isn’t easy — especially when the job titles sound so similar and the roles have a good amount of overlap. Data analytics, data engineering, data science, machine learning engineering, software engineering — while there are definitely similarities, the day-to-day work and the type of person who thrives in each role couldn’t be more different. And if you choose wrong, you could end up spending tons of time and money building skills you’ll never use, or worse, be stuck in a job that makes you miserable.

After years working across these roles myself and mentoring people breaking into tech, I’ve seen way too many folks realize six months into their new job that they picked the wrong path.

So, today I’m breaking down exactly what you need to know about each of these careers to make the right choice for you. Not just the job descriptions, but the real, day-to-day work. What kinds of problems will you actually solve? What skills do you really need? And most importantly — which role fits YOUR strengths and personality?

We’ll cover:

• The surprising differences between data science, analytics, and engineering roles
• The pros and cons you might not know about from the job description
• And how much money you can expect to make

By the end of this post, you’ll have a super clear picture of which path is right for you. No more confusion or doubt.

Let’s get to it.

Role Overviews

Data Analyst

Let’s start with Data Analysts. You know those articles that say ‘millennials are killing the napkin industry’ or ‘this new product increased sales by 47%’? There’s probably a Data Analyst behind those insights, digging through the numbers to figure out what’s going on in the business and why.

You can think of Data Analysts as sort of, business detectives. When the marketing team is trying to figure out why their latest campaign flopped, or the product team needs to understand why users are dropping off at a specific feature — Analysts are the ones who figure it out.

Here’s what their day might look like: The head of marketing has an urgent question: ‘Our Instagram ads are costing twice as much as last month. What’s going on?’ The Analyst dives in, pulling data from multiple sources, writing SQL queries, and making visualizations to piece together the story. After some investigation, they discover that the team has been targeting the wrong audience segment — and a simple fix that ends up saving the company thousands of dollars on ad spend.

If that sounds fun to you, let’s break down what you actually need to succeed in this role.

First, the technical stuff:

• SQL is number one. You’ll be using this all day, every day.
• Visualization tools like Tableau or Power BI will be a big part of your work.
• You will likely need to be comfortable with Excel for data wrangling and simple visualizations.
• Basic Python or R will be needed for more complex analyses.
• And you’ll need enough stats knowledge to know if your findings are legit or just random noise.

But what can often be missed is that the most important skills for really great Analysts aren’t technical at all. You need:

• The ability to translate vague business questions into concrete analyses.
• Communication skills to explain complex findings to stakeholders.
• Business intuition to know which questions actually matter.
• And critically, the ability to tell a story with data.

In terms of compensation, Data Analysts in the US have a median salary of around $106k total compensation. Not bad for one of the more accessible paths into tech.

Now for some pros and cons.

The good stuff about being an Analyst :

• It’s one of the easier tech roles to break into. No PhD required.
• The skills of a Data Analyst are valuable across industries like healthcare, finance, marketing, and tech, so there are more opportunities outside of traditional tech companies (which means more job opportunities!).
• You get to see the impact of your work quickly. When you find an insight that saves the company money or improves a product, people notice.
• The skills you learn set you up well to transition into data science or even roles like product management later on if you want to.
• If you enjoy both solving puzzles and storytelling, this role lets you do both.

But it’s not all sunshine and rainbows :

• You’ll spend more time cleaning messy data than you probably expect, which can be tedious.
• Analysts typically work with structured data and predefined questions, which can feel repetitive or less exploratory compared to other roles.
• The role generally requires less coding and technical expertise, so if you’re someone who wants to be learning new techniques and tools, it may not be a fit.
• The pay, while good, typically caps out lower than engineering roles.
• And here’s the biggest one, in my opinion: You will get a lot more ad-hoc requests from stakeholders. It can end up being a very reactionary role, where you’re just handling what gets put in front of you vs. being able to strategically think about the kinds of projects that you think are the most interesting or impactful, or what is best for your career.

So, if you like the idea of working with data but want more autonomy and higher technical challenges, you might want to consider our next role: Data Engineering. These are the folks who build the infrastructure that makes an Analyst’s job possible. Let me show you what I mean…

Data Engineer

Imagine Netflix. Every second, millions of users are streaming shows, rating content, and searching for titles. All of that creates a ton of data. Data Engineers are the ones who build the systems that capture all this information, clean it up, and make sure it ends up in the right place for others to use.

Here’s what that actually looks like: Let’s say Netflix wants to build a new recommendation system. While the Data Scientists or Machine Learning Engineers are brainstorming algorithms, the Data Engineer has to figure out how to:

• Build systems to ingest and process millions of real-time events per second from user devices.
• Design data models that can track user identity and behavior across shared accounts.
• Ensure data is properly indexed and cached for fast access by recommendation systems.
• Set up monitoring to catch data quality issues and pipeline failures before users are affected.
• Implement validation rules and checks to ensure data accuracy and completeness at each step.

If this sounds like a technical puzzle you’d enjoy solving, here’s what you need in your toolkit :

• Strong Python, Java, and/or Scala skills for building data pipelines (but mostly Python)
• Deep knowledge of databases (both SQL and NoSQL)
• Experience with data processing tools like Airflow, Spark and Hadoop
• Cloud platform expertise
• Experience with streaming technologies like Kafka and Kinesis
• And most importantly, a keen attention to detail

Money-wise, Data Engineers in the US typically make around $155k in total compensation.

Let’s talk about what makes this role great :

• You get to solve complex technical challenges that actually matter.
• There’s a lot of satisfaction that comes from building systems that work flawlessly.
• The field keeps evolving with new tools and challenges, so you’re always learning.
• Strong job security since every tech company needs data infrastructure.

But of course, there are some downsides :

• Your best work often goes unnoticed — nobody comments when the data pipeline works perfectly.
• You might feel like you’re always playing a supporting role to Data Scientists or MLEs who build on top of your foundational work.
• The on-call responsibilities can be demanding when data pipelines break.
• And at bigger companies, you might get stuck maintaining legacy systems instead of building new ones.

So, if you love building reliable systems and don’t need the spotlight, this could be your perfect role. The satisfaction comes from knowing that Data Scientists, Analysts, and basically everyone else in the company relies on your work to do their jobs.

Speaking of Data Scientists…

Data Scientists

Data Science is probably the most confusing role of all, because nobody can seem to agree on what they actually do across different companies. But here’s the general idea.

Data Scientists tackle ambiguous problems that can’t be solved with simple analysis. They build statistical models to predict customer behavior, design experiments to measure causality, and develop algorithms to automate complex decisions. While analysts might tell you what happened last quarter, Data Scientists build models to predict what will happen next quarter. But before any sophisticated modeling happens, there’s usually weeks of cleaning messy data, defining what success actually means, and convincing stakeholders that the new approach is worth the investment.

Let me give you a real example: Imagine you’re a Data Scientist at DoorDash, and you’re tasked with predicting delivery times. You have lots of historical data to work with so this should be pretty straightforward, right? Well…

• First, you need to figure out what data actually matters. Is it traffic? Weather? The restaurant’s busy hours? What mood the drivers are in?
• Then you discover half your historical data is missing some key feature and you need to get creative about how to handle that.
• You build a model and it works great… except during rain, or sports events, and so on.
• Just when you think you’ve got it figured out and you’re done, someone asks “but can we explain this to our restaurant partners?”
• And, depending on your team, you might also need to deploy this model to production and make sure it behaves as expected once real users are relying on it.

This brings us to what you actually need to succeed :

• Python or R skills for everything from data cleaning to modeling (but again, typically Python).
• Strong statistics background.
• Machine learning knowledge for building predictive models.
• SQL for getting your hands on the data you need.
• And crucially, the ability to explain complex concepts to people from non-technical to PhDs.

In terms of compensation, Data Scientists in the US typically make around $168k total comp.

Now, what makes this role awesome :

• Despite recent layoffs and fears around AI, there continues to be strong global demand for Data Scientists.
• Data Scientists typically earn more than Data Analysts and often on par with Data Engineers and Machine Learning Engineers.
• There’s a lot of flexibility in the kinds of roles available and industries to work in. The role can vary significantly, with tasks ranging from predictive modeling to supporting analytics for business teams, and there are jobs in industries from finance to healthcare to tech.
• The field is constantly evolving, so you’re always learning.
• And when your models work, they can have a measurable impact, which can be really satisfying.

But of course, there are challenges :

• The barrier to entry is high, with most roles requiring an advanced degree and/or significant experience.
• Like with Analytics, there’s lots of time spent on the more mundane data cleaning side of things.
• Similarly, new folks expect to work on cutting-edge AI but find themselves performing more routine analysis or visualization tasks.
• The role is super ambiguous — what makes a “good” Data Scientist varies wildly between companies. You need to be ok with figuring out your niche in a company and learning a lot on your own.
• And finally, the career progression is less defined than in engineering roles. What makes someone a “Senior” Data Scientist can vary significantly from company to company.

Data science is ideal for those who thrive in multidisciplinary and exploratory environments. It’s not the best choice if you want to work on more cutting edge technical challenges, since that’s not a major part of many roles. If that bums you out, the next role might be a better fit.

Machine Learning Engineer

You can think of it this way: if Data Scientists are like researchers testing new recipes in a kitchen, Machine Learning Engineers are the ones who figure out how to serve those recipes to thousands of people every day with a meal delivery service.

Here’s a scenario: You’re a Machine Learning Engineer at Uber, and the data science team has built this amazing model that predicts surge pricing. Sounds great! But now you have to:

• Make it return predictions in milliseconds, not seconds.
• Ensure it works 24/7 without crashing.
• Handle millions of requests during peak hours.
• Update the model without disrupting the service.
• Monitor it and make sure nothing unexpected happens.

So, Machine Learning Engineers are kind of a bridge between data science and software engineering. To thrive in this role, you need a pretty unique skill set :

• Strong software engineering fundamentals (design patterns, testing, CI/CD)
• Deep experience with ML frameworks and tools like PyTorch, Hugging Face, and MLflow.
• Systems design skills to handle scale (think distributed training and model serving).
• Practical ML expertise to optimize model performance and troubleshoot issues.
• Production engineering skills for things like monitoring, debugging, and performance tuning.

Now one of the nice things about this role is that Machine Learning Engineers typically make around $250k in total comp in the US. So that’s pretty cool.

Some other things that make this role awesome :

• You often get to work with cutting-edge tech.
• The impact is super clear — your systems directly affect users in a measurable way.
• The role has a somewhat clearer career progression than data science and data analytics since they are in the engineering family.
• It’s growing fast right now. There are a lot of jobs out there for highly-skilled Machine Learning Engineers.
• And you get to solve some really interesting technical puzzles.

But of course, there are challenges :

• Like data science, machine learning engineering has higher entry barriers since it requires a strong background in both machine learning and software engineering, often with advanced degrees or significant experience.
• The field moves incredibly fast, so you’ll need to enjoy continuous self-study to keep up (which might be a pro for some).
• Production issues can be stressful.
• Often, ML models just don’t work like you want them to. It might be a data quality issue, or just a really hard problem to solve, and there won’t be a way to get the accuracy you want. You have to be ok with ambiguous problems with no clear right answer.

The work of a Machine Learning Engineer is less about inventing new algorithms and more about making existing ones work reliably at scale. This career is ideal for folks who like engineering, but are ok with or even energized by working on ambiguous problems that might not have a correct answer.

If you’re listening to this and thinking you’d rather work on more traditional software systems where things are a bit more predictable, our next role might be perfect for you. Let’s talk about Software Engineering, where your code either works or it doesn’t (mostly).

Software Engineer

While everyone is talking about AI and machine learning, software engineers are quietly building pretty much everything you use every day: the apps on your phone, the websites you browse, even the systems that keep planes in the air.

Let me walk you through what software engineering actually looks like. Say you’re a Software Engineer at Spotify working on a new podcast feature. Your day (or month) might involve:

• Starting with a product spec and architecture discussion about how to handle podcast clip sharing.
• Breaking down the technical requirements: media timing APIs, cross-platform compatibility, sharing protocols.
• Writing code and tests for the feature while considering edge cases and performance.
• Collaborating with other teams on API design and integration points.
• Deploying changes gradually, monitoring for errors, and iterating based on user feedback.

What makes software engineering both challenging and satisfying is that you’re building systems that impact millions of users. Here’s what you need in your toolkit for the job :

• Strong coding fundamentals and experience with modern languages (Python, JavaScript, TypeScript, Go).
• Practical knowledge of data structures, algorithms, and system design.
• Experience with cloud services, containers, and deployment tools.
• Understanding of testing practices, CI/CD pipelines, and monitoring.
• Ability to write maintainable code and effectively debug production issues.

The median total comp for SWEs is $182k in the US. Not as high as ML Engineers, but still pretty sweet.

Let’s talk about what makes this role great :

• Your work is tangible — you build things that actually exist.
• There is a clear feedback loop — your code either works or it doesn’t.
• SWEs are essential in almost every industry, making this a highly versatile career path where you can choose to focus on lots of different areas.
• Similarly, you can focus on many different specializations based on what you like best, from front-end, back-end, full-stack, DevOps, or even things like blockchain or augmented reality.
• And compared to the data and ML jobs we’ve discussed, it’s easier to get in the door. While degrees can help, many SWEs start through bootcamps or self-study, making the field more accessible.

But of course, there are trade-offs :

• It can be a bit more repetitive and less imaginative than data or ML roles.
• Like machine learning engineering, you will need to continuously be learning new tools and frameworks (though again, this could be a pro for the more intellectually curious).
• Also like machine learning engineering, you will be working with production systems that have a lot of pressure if things go wrong. You’ll likely be on-call from time-to-time and have fire drills when things break.

So, while data roles are amazing for solving complex analytical problems, software engineering is perfect if you love building concrete solutions and seeing immediate results.

How to Choose

Now that we’ve covered each role, let’s talk about how to actually choose between them. In my opinion, they all sound pretty fun, but picking the wrong one can mean months or years of building skills you won’t use, or worse, being stuck in a role that drains your energy every day.

If you’re still feeling unsure, I’ve created a detailed quiz that walks you through all these factors we just discussed. It asks targeted questions about your preferences, skills, and working style to help match you with the right role. It’s free and takes about five minutes to complete.

In the meantime, let’s break this down some more with some real examples to help you decide.

Key Decision Factors

First up: Do you prefer building or discovering? This is huge, because it completely changes how your day-to-day work feels.

Imagine you’re working on a project to improve user engagement with a new collaboration feature.

• As a Software Engineer, you’re the one actually building the new feature. When it’s deployed, you’ll see people use it immediately. So, pretty concrete results.
• As a Data Engineer, you’re building pipelines to track how people use the feature. Your success is measured in how reliable and comprehensive the data flow is.
• As a Machine Learning Engineer, you’re evaluating the matching algorithm’s performance and trying to figure out why latency is worse for some groups, or why the algorithm performs poorly on a subset of the users.
• As a Data Scientist, you’re diving into user behavior patterns, testing hypotheses about what makes people engage more. Some days you find amazing insights, other days you realize your theory was completely wrong.
• And if you’re a Data Analyst, you’re focused on answering specific questions: Which user groups love the feature? Where are people dropping off?

So as you can see, some roles give you the satisfaction of building concrete things, while others are more about exploration and discovery.

Next up: How much structure do you want in your role?

Think about it like this:

• Software Engineers and Data Engineers often have clear tasks: “Build this feature” or “Create this pipeline.” You know exactly what success looks like.
• But for Data Scientists? Their tasks are more like “Figure out why users are leaving” or “Make our recommendations better.” This is way more ambiguous.
• Machine Learning Engineers sit somewhere in the middle — you have concrete engineering tasks but with the uncertainty of ML thrown in for fun.
• Data Analysts might get specific questions but often have to figure out the right approach themselves.

Now let’s talk about skills , because this is where people often get tripped up.

Here’s a scenario: You love math and are excited about AI. Naturally, you think “I should become a Data Scientist!” But hold up — do you actually enjoy:

• Explaining complex concepts to non-technical people?
• Writing detailed analysis documents?
• Dealing with ambiguous business problems?

If not, you might be happier as a Machine Learning Engineer, where you can work with ML but focus more on the engineering side.

Or maybe you love working with data but hate the idea of being on-call for production systems. In that case, data analytics or data science might be better fits than data engineering or software engineering.

Let’s talk about barriers to entry too, because this matters in the real world:

• Want to get started quickly? Data analytics is your fastest path. You can learn the core tools (SQL, Excel, visualization) in a few months.
• Software engineering has a steeper learning curve, but bootcamps and self-study can get you there without an advanced degree.
• Data engineering roles usually want some software engineering experience first, but you can get in without advanced degrees.
• But, data science and machine learning engineering often need advanced degrees or significant experience. Not always (if you’re exceptional), but often.

Ultimately, there are tradeoffs to all of these options, but your success will come from aligning your career choice with what excites you most, and playing to your strengths.

Common Overlaps and Transitions

One thing to keep in mind is that the lines between roles can be blurry, and it’s common to find significant overlap between data and engineering roles.

Picture a startup with 50 people. Their ‘Data Scientist’ is probably doing analytics on Monday, training ML models on Tuesday, and figuring out data pipelines on Wednesday. Meanwhile, their Software Engineer is learning MLOps because someone needs to deploy those models.

Even at larger companies, the boundaries blur:

• Software Engineers may work with ML tooling and model serving infrastructure.
• Data Scientists sometimes write production code and own end-to-end ML pipelines.
• Data Engineers may architect feature stores and experiment tracking systems.
• And everyone gets involved with reliability and monitoring.

This overlap isn’t a bad thing — it actually creates natural pathways to transition between roles. So, if you do end up picking a role that’s not perfect for you, all is not lost. Let me show you some common routes people take:

Data Analyst → Data Scientist:

• Many Data Analysts use their foundational skills in querying and visualization to move into data science.
• Adding statistical analysis, machine learning, and more rigorous programming to their toolkit helps bridge the gap.
• But, transitioning often requires additional education, such as a master’s degree or certifications, and hands-on experience with data science projects.

SWE → Data Engineer or → MLE:

• Software engineers with a strong background in building systems often transition to data engineering, where they apply their coding and system design expertise to data infrastructure.
• Or, they may acquire knowledge of machine learning frameworks and algorithms to enable switching into MLE roles.

Data Scientist → MLE:

• This was my path!
• Data Scientists with a focus on building and deploying models may transition into machine learning engineering by strengthening their software engineering and MLOps skills.
• Gaining experience with ML pipelines and production environments is key here.

So, career transitions often require targeted skill-building, but they are achievable with some deliberate effort and strategic project choices.

If you’re just starting out, one good strategy might be to consider roles that offer broad exposure, like data analytics or software engineering, to build a versatile foundation. Similarly, if you’re looking to pivot, focus on acquiring complementary skills that bridge the gap between your current role and your target role.

Remember, your career is a journey, not a fixed destination. Focus on building skills, solving meaningful problems, and staying open to opportunities as they come.

Choosing between these roles — data analytics, data engineering, data science, machine learning engineering, and software engineering — can feel daunting, but the right choice is the one that aligns with your passions, strengths, and long-term goals.

— — —

If you’re still not feeling 100% sure, check out this free quiz to help you figure out which role is the best fit for you!

Here are some other ways I can help:

• If you’re interested in becoming a Data Scientist, I have a really comprehensive video on YouTube with a complete roadmap on how to go from an absolute beginner to your first job.
• Or, I have a free 80+ page e-book with the same roadmap info, plus links to learning resources, checklists, and more!
• If you’d like to chat 1:1, you can book a call with me here.