24 July 2025
How We See Data Science in the AI Era
What happens when AI takes over the tasks once reserved for junior analysts? In this conversation between a mentor and their interns, we explore how to start a real career in data science when the easy work is gone. Through skill maps, career tracks, and reflections on what AI still can’t do, this is a guide for those who want to thrive by choosing hard, meaningful problems—and learning to connect data with context.
Piyorot Piyachan
CEO of Mutrack
When I sit down with new interns or junior hires on our data team, I no longer begin by explaining what we do at Mutrack or what tools we use. I start with a quiet truth: the career path you thought you were on might not exist anymore.
It’s not your fault. Most universities still teach data analysis like it’s 2015. They hand you Python, R, some linear regression, a few business cases, and a capstone project that ends with a dashboard. But that path—the one that starts with cleaning data, writing queries, building reports, and slowly moving up to machine learning and strategic work—has already been flattened by artificial intelligence. Entry-level jobs, as we knew them, are disappearing. AI now handles a large portion of what junior analysts used to do. What we called “easy tasks”—routine querying, basic modeling, and even interpreting simple trends—are increasingly managed by tools that require no analyst at all.
That’s why I believe every aspiring data professional needs to confront a hard question early: If AI can do what I thought I was going to do, then what should I be preparing for?
In the age of artificial intelligence, data is no longer just information—it’s intuition made visible. What you see here isn’t just form, it’s pattern, uncertainty, possibility. 🧤
This isn’t a crisis. It’s a call to clarity. There is still incredible demand for data talent. But that demand has shifted. We no longer need people who can run reports. We need people who can frame problems, challenge assumptions, and translate messy operational questions into structured insights. We need people who understand both the business and the data. And above all, we need people who know how to work on what MIT economist Daron Acemoglu calls “hard tasks”—the contextual, strategic, high-ambiguity work that AI still struggles with. The easy stuff is gone. The hard stuff is where you must begin.
You Need a Map, Updated One
When I talk to my interns about this, I usually pull out a chart—a skill map I built for our data analysts and scientists at Mutrack. It’s not a checklist. It’s a map of possibilities. Across this map are the real competencies we rely on in our product and operations: industry knowledge, market research, product thinking, statistics, machine learning, web development, DevOps, data pipelines, presentation skills, and more. No one masters them all. But everyone must pick a corner and go deep.
This is our skill map for data professionals in the AI era—built not as a checklist, but as a compass. No one masters it all. But everyone must choose a direction, go deep, and make it matter. 🎩
We start with Industry Knowledge. This isn’t just knowing what the business does. It’s about immersing yourself in the ecosystem. In our case, that means hospitals. How does a patient journey unfold from arrival to discharge? What bottlenecks frustrate staff and delay care? What unspoken trade-offs do hospital administrators make every day? Without this context, your data work is untethered—like diagnosing a patient without understanding the hospital they're in.
Then there’s Market Research. This is your radar. It’s your ability to look outside the walls of your own data warehouse and understand what’s happening in the industry. What trends are emerging? What competitors are moving faster? What macro shifts—like policy or technology—might affect decision-making? Good analysts aren’t just number crunchers. They’re context seekers.
Product Design and Development follows. At first glance, this doesn’t look like a data skill. But if your insight doesn’t lead to action, what was it for? Understanding product design means understanding constraints. What makes a product usable? What makes a feature stick? If your chart can inform a better design, or your model can support a new feature, you’re no longer a back-office analyst—you’re a builder.
Now we move to the technical middle of the map. Programming is your toolset. It’s how you automate, build, test, and scale your ideas. It’s not about how fancy your code is—it’s about how effective and repeatable your solutions are. If you know how to code, you know how to ship.
Statistics is where your judgment begins. You need more than averages. You need to understand distributions, biases, confounding variables, and causality. Statistical thinking lets you question your own output. It keeps you honest.
From there we expand to Machine Learning. It’s not about fitting neural networks for the sake of it. It’s about when to use learning systems, how to evaluate models, and when to say, “We don’t need AI—we need a simple rule.” Machine learning is powerful, but restraint is just as important.
Data Preparation is the unglamorous heart of analysis. Every dataset tells a different story depending on how it’s cleaned, filtered, and shaped. Knowing how to wrangle messy data is still essential. It’s your first signal of whether the work ahead is meaningful or misleading.
Modeling is about translating questions into structured forms. It’s about building abstractions, testing them, and refining them over time. Whether you’re forecasting, segmenting, or running simulations, modeling is how you make ideas concrete.
Data Visualization is how you guide attention. It’s how you say, “Look here. This is what matters.” Great data people don’t make pretty charts. They design conversations. They help others see what they see.
Further into systems, we arrive at DevOps and Database Management. These are often overlooked by analysts but essential for anyone working in production environments. If you want your insight to live inside a system that runs every day, you must understand infrastructure. How do we deploy updates? Where does the data live? What happens when it grows 10x?
Big Data Infrastructure extends this further. When you go from thousands to millions of rows, or when data comes from real-time sources like sensors or applications, you must know how to scale. Big data isn’t about volume—it’s about design decisions.
Data Streaming and Pipelines follow. This is about movement—how data flows from one place to another in real time. If your system relies on up-to-the-second information, like our dispatch engine for hospital porters, then building reliable, monitored, fault-tolerant pipelines is not just nice to have—it’s life-critical.
CI/CD, or Continuous Integration and Deployment, brings discipline to all of this. It’s how we test, verify, and release improvements with confidence. It’s also how we prevent breaking everything when we’re just trying to fix one thing. Analysts who understand CI/CD can participate in engineering conversations—and ship faster.
Last but never least, Presentation and Storytelling. This is the multiplier. It’s not about sounding smart. It’s about making others smarter. When you can tell the story behind the data—what matters, what changed, what we should do next—you become irreplaceable. Data becomes strategy.
After we cover the map, we step back. It’s too much to master all at once. So I offer a simpler lens: What kind of data person do you want to become?
At Mutrack, we describe three archetypes: Product-Focused, Data-Focused, and Ops-Focused. Each is a legitimate, defensible path. Each creates real value. And more importantly, each comes with its own rhythm of work, source of fulfillment, and path toward mastery. Let me walk you through what each track truly looks and feels like when you live it for real.
You Love Building Great Products
If you’re Product-Focused, you live in the intersection of customer insight, product evolution, and data interpretation. You don’t wait for a Jira ticket. You initiate conversations. You ask, "Why are we building this? What’s the real need?"
You might find yourself sitting with a nurse during a hospital visit, watching how they juggle screens, forms, and patient needs, just to discover a 3-second friction that no metric captured. You are part analyst, part UX whisperer, part mini-PM. You love figuring out not just what’s broken—but what’s missing. You use data as your compass, not just your evidence. Your presentations change roadmaps. Your questions shape features.
You are trusted not because you have all the answers, but because you ask the most clarifying ones. Over time, you become the person people turn to when they want to build things that actually matter.
You Enjoy Coding and Digging
If you’re Data-Focused, you go deep. You’re the kind of person who loses track of time trying to answer, "Why is this pattern showing up?" You build models not for show, but to test, learn, and iterate. You're skeptical of assumptions—even your own.
You enjoy the elegance of a well-written function and the intellectual challenge of a multivariate A/B test. You obsess over statistical rigor. You care about cleaning the data as much as analyzing it. And yet, you’re not trapped in theory. You want your work to influence decisions, shape strategy, and unlock value. In your hands, a messy dataset becomes a story, a prototype, a lever. You learn to zoom in with precision and zoom out with wisdom.
People may not always understand your equations—but they understand the clarity your work brings. That’s your power.
You Dream to Make Big Systems Run
If you’re Ops-Focused, you are the engineer behind the curtain. You take pride in making complex systems look simple on the outside. You love when a job dispatch gets sent to the right person in the right place without anyone noticing the orchestration underneath.
You think about latency, failovers, monitoring, and scale—not because it’s glamorous, but because it’s the kind of work that must be done right the first time. You work with databases, pipelines, deployment pipelines, and edge cases. You thrive in real-time environments. When systems go down, you’re the one people call. But more importantly, when things run smoothly, it’s because of you.
You find elegance in efficiency, strength in reliability, and pride in being invisible but essential. You might not be center stage, but you’re the one holding the rigging ropes so everyone else can perform. And in your quiet mastery, you create the conditions for everything else to thrive.
Navigating Data Science in the AI Era
Each track has its beauty. But what unites them all is a shared truth: in this new world, your career starts not with easy tasks, but with meaningful ones. You must choose hard problems early. You must get used to ambiguity. You must learn to lead, not just execute.
When people say “data is the new oil,” they forget that oil has no value unless it’s refined, transported, and used. The same is true for data. We don’t need people who just extract it. We need people who make it matter.
So if you’re about to work in the data-rich world, here’s my advice: Don’t look for the ladder. It’s gone. Look for a wall worth climbing. Choose your grip carefully. And pull others up with you.
The Human Work Ahead
And if we circle back to where we started—to Daron Acemoglu’s argument about easy and hard tasks—the lesson becomes clearer. In the age of AI, it’s no longer enough to master one discipline. Knowing how to write code or fit models won’t protect you. The tasks that survive—those hard, ambiguous, deeply human tasks—demand more than skill. They demand awareness.
In this increasingly complex world, filled with data streams and soon, super-intelligent agents, it is still possible for humans—just like you and me—to not only survive, but to thrive. Not because we process faster, or calculate better. But because we understand people. Because we see nuance. Because we recognize contradiction and possibility where AI sees only patterns. We bring context. We bring judgment. We bring purpose.
The real work now is connecting the invisible dots—across teams, across tools, across silos. And we do that through our own lens, shaped by experience, empathy, and conviction. That is what makes a great data professional today.
So choose your track. Pick a hard problem. Embrace complexity. Stay close to reality. Keep asking questions. And above all, don’t forget the human behind the data. That’s where the future still belongs to us.
Continue Reading
17 July 2025
Piyorot Piyachan
CEO of Mutrack
10 July 2025
Piyorot Piyachan
CEO of Mutrack
3 July 2025
Piyorot Piyachan
CEO of Mutrack