The Unexpected AI Skills I Gained from a PhD in Analytical Chemistry

From multivariate calibration to machine learning pipelines — the scientific mindset travels farther than I thought (AI in Chemistry)

When I was deep into my PhD in Analytical Chemistry, surrounded by chromatograms, fluorescence signals, and endless spreadsheets of spectral data, I never imagined that years later I’d be applying similar thinking patterns to train AI models.

But here I am — learn artificial intelligence with the same scientific rigor I once applied to the detection of polycyclic aromatic hydrocarbons (PAHs) in complex matrices.

This post isn’t just a story about changing fields. It’s about how a scientific background — specifically in Analytical Chemistry — can be a surprisingly solid foundation for AI. If you’re a scientist looking beyond the lab bench, or someone in tech curious about what science brings to the table, read on.

The quiet intersection: chemometrics and early AI in Chemistry

During my doctoral research, I worked on detecting PAHs using chemometrics — a form of multivariate analysis that’s incredibly similar to what we now call classical machine learning. Techniques like Principal Component Analysis (PCA), Partial Least Squares Regression (PLSR), and even Artificial Neural Networks (ANNs) were essential tools in my daily work.

Back then, we didn’t call it “AI” (AI in Chemistry) we called it modeling, data fitting, or just “making sense of the noise.” But in hindsight, I was already doing machine learning — just within the context of spectroscopy and environmental analysis.

We were solving real problems: low detection limits, interferences, method validation, repeatability. It required clean thinking, structured experimentation, and data-first reasoning. These are not just chemistry skills. They’re transferable, powerful, and directly applicable to the world of AI.

From curiosity to career shift

Like many scientists, I didn’t leave the lab because I hated science. I left because I was curious about the world beyond the PhD.

What started as passive reading on data science blogs turned into nights spent watching online courses, downloading Kaggle datasets, and testing small Python scripts. I didn’t make the leap immediately. I needed time — not just to learn new tools, but to reframe how I saw myself.

And here’s what helped: realizing that the mindset I developed in chemistry was still incredibly valuable.

What Analytical Chemistry teaches you about AI

I used to believe that switching into AI meant starting from zero.

I was wrong.

Here are the top lessons from my scientific background that now shape how I learn and apply AI:

1. The scientific method is the ultimate debugging tool

When you’ve spent years setting up experiments, testing one variable at a time, documenting every possible error source — you develop a kind of mental discipline that’s rare in fast-paced software culture.

That discipline helps in AI too:

• When your model overfits: you go back to basics.
• When your metrics don’t improve: you question assumptions.
• When your pipeline fails: you isolate and test each component.

You’re not just coding — you’re experimenting.

2. Thinking in dimensions

Spectral data is inherently multivariate. Each sample is a vector of absorbance values, intensities, or emission peaks.

So when I started working with feature matrices in machine learning, it felt… familiar. Instead of wavelengths, I now deal with pixels or token embeddings. But the intuition for multidimensional data is the same.

3. Data integrity is everything

In Analytical Chemistry, garbage in = garbage out is not just a phrase — it’s a daily battle. You don’t trust a result until the controls check out, the baseline is clean, and the method is validated.

In AI, we often forget this. But my chemistry mindset won’t let me.

Before training a model, I scrutinize the data. I look for drift. I question the labels. I inspect outliers. Because I know — painfully well — how small flaws in input data can destroy the entire analysis.

4. Communicating uncertainty

Another transferable skill? Reporting results with uncertainty.

Chemists don’t just give numbers. We give ranges, confidence intervals, detection limits. We know how to interpret error and talk about limitations.

In AI, that becomes an asset — especially in sensitive fields like healthcare or education, where overconfidence can have real consequences.

Small AI steps, real-world value

Today, I’m working on small applied projects in AI. Some involve OCR (optical character recognition). Others involve image detection pipelines. They’re far from cutting-edge research — but they’re real.

And that’s what excites me: not chasing hype, but applying AI responsibly and effectively in domains I care about — like education and health.

These are fields where interdisciplinary thinking is desperately needed. Where data is messy, stakeholders are non-technical, and the margin for error is small. In other words — exactly the kind of environment analytical chemists thrive in.

Interdisciplinarity is not a buzzword — it’s a career

There’s a strange belief in tech that you need to have a “pure” background to succeed. That you need to be a software engineer from day one, or a data scientist with only one focus.

But the real value often lies in the intersection of disciplines.

My chemistry PhD taught me how to structure problems. AI gives me the tools to scale those solutions. Together, they create a profile that’s hard to find — and even harder to replace.

So if you’re reading this and wondering whether your scientific training is still relevant in the AI world, let me be clear:

It absolutely is.

Not only relevant — it’s needed.

What’s next?

My journey into AI is far from over. I’m still learning every week. Still writing ugly Python scripts. Still breaking things and reading documentation at 1AM.

But I now see this path not as a detour — but as a continuation of the same mindset that got me through a PhD:

• Ask good questions.
• Follow the data.
• Think critically.
• Communicate clearly.
• Never stop learning.

Final thoughts (and a quiet call to action)

If you’re a scientist wondering where to go next — don’t discard what you already know.

If you’re in AI and hiring for your team — don’t overlook candidates with PhDs in fields like chemistry or biology.

And if you’re like me — somewhere in between disciplines, learning as you go — know that you’re not alone.

I’m happy to connect with others walking this path. Let’s learn together.