Starting with Math (And Questioning Everything)#
I was a math major at UCSB, spending way too many hours on proofs and abstract concepts that felt completely disconnected from reality. Honestly, I kept wondering “when will I ever use this?” Real analysis seemed like pure torture, and I couldn’t see how any of it would help me get a job.
Turns out I was wrong about that, but it took me a while to figure out why. The math itself wasn’t the point—it was more about training my brain to think systematically. Now when I’m staring at messy data that makes no sense, I automatically start breaking it down into smaller pieces, looking for patterns. I guess all those proofs taught me something after all.
My First Reality Check: Data Science is Mostly Not Science#
My first real project was trying to predict poverty rates using census and satellite data. I was so naive—I thought it would be like a homework assignment: download data, run some sklearn models, get perfect results. Reality hit hard.
The data was an absolute disaster. Missing values everywhere, columns with cryptic names, and when I finally got something to run, the results made zero sense. I spent probably 80% of my time just trying to figure out what the data even meant. Turns out understanding the problem is way harder (and more important) than knowing which algorithm to use. Still learning this lesson honestly.
The AI Thing is Making Me Anxious (But Also Excited?)#
Okay, real talk: when ChatGPT came out, I kind of freaked out. I’d just spent months learning Python, and suddenly this thing could write code better than me. Made me question everything I was doing.
But after using AI tools for a while, I’m starting to think about it differently:
It’s Not About the Code Anymore#
I used to think being good at data science meant memorizing pandas functions and sklearn syntax. But ChatGPT can already do that stuff. What it can’t do is figure out that the problem you’re trying to solve isn’t actually a machine learning problem—it’s an operations research problem, or a communication problem, or sometimes not a problem worth solving at all.
Like with my Michigan Medicine project, I had to figure out that pharmacists weren’t using the better system because of trust issues, not technical issues. No AI would have figured that out from the data alone.
I’m Trying to Learn How to Learn#
This might sound obvious, but I’m realizing that specific skills become outdated so fast now. The transformer stuff I learned last year? Already feels old. So I’m trying to focus more on understanding underlying principles rather than memorizing specific implementations.
My math background actually helps here. When new AI architectures come out, I can usually understand the core ideas because I’m comfortable with the math. But I still feel like I’m constantly playing catch-up.
Mistakes I’m Still Learning From#
Thinking Fancier = Better#
I definitely went through a phase where I thought complex algorithms were always better. Deep learning everything! If it wasn’t neural networks, it wasn’t worth doing.
But then I did this DEA analysis on renewable energy in Iran, and the simple linear programming approach gave much clearer insights than any fancy model could have. Sometimes the people you’re presenting to just want clear answers, not black-box predictions.
Forgetting That People Matter#
This one’s embarrassing, but I used to think if my analysis was mathematically correct, that was enough. Like with the Michigan Medicine project—the data clearly showed the new method was better, so why weren’t people using it?
Turns out it doesn’t matter how good your solution is if people don’t trust it or find it too complicated. I’m still figuring out how to balance technical optimization with human factors.
Not Asking “So What?” Enough#
I got really good at building models and generating impressive-looking results. But I wasn’t always great at stepping back and asking whether anyone actually cared about what I was measuring. I’m trying to start every project now by understanding what decision this analysis is supposed to support.
What I’m Hungry to Learn Next#
Going Deeper into Operations Research#
I’ve gotten a taste of optimization through my projects, but I feel like I’m just scratching the surface. There’s so much I don’t know about metaheuristics, stochastic programming, multi-objective optimization. I want to understand not just how to use these tools, but when and why they work.
Getting Better at System Design#
Most of my projects have been one-off analyses, but I’m realizing that real impact comes from building systems that people can actually use. I need to learn more about software engineering, databases, cloud architectures—all the stuff that turns a notebook into something production-ready.
Understanding Different Domains#
I’ve done projects in healthcare, energy, and poverty analysis, but I feel like I barely understand any of these fields deeply. I want to pick one or two domains and really dive in—understand the business context, the constraints, the politics. Maybe healthcare operations or sustainability analytics.
Working with AI, Not Against It#
I’m trying to figure out how to use AI tools to accelerate my learning rather than replace my thinking. Like using ChatGPT to explain concepts quickly, but still working through the math myself to really understand it. Or using Copilot to write boilerplate code faster so I can focus on the interesting parts.
Things I Wish Someone Had Told Me#
Most of Your Time Won’t Be Modeling#
This sounds obvious now, but I thought data science would be mostly building cool algorithms. In reality, I spend way more time figuring out what the actual problem is, cleaning messy data, and trying to explain results to people who just want simple answers.
Start Simple, Then Get Fancy#
I used to think using the most advanced technique made me look smarter. But starting with linear regression or basic statistics often gives you insights that more complex models miss. Plus, it’s a lot easier to explain why your simple model is wrong than why your neural network is wrong.
Your Code Will Become Obsolete, But Problem-Solving Won’t#
The pandas syntax I memorized is already being replaced by new tools. But the ability to look at a messy situation and figure out what questions to ask? That’s not going away.
What I’m Still Figuring Out#
How to Stay Relevant When Everything Changes So Fast#
Honestly, this keeps me up at night sometimes. New AI models come out every month, new frameworks replace old ones, and I feel like I’m always behind. I’m trying to focus on understanding principles rather than memorizing specifics, but it’s hard to know if that’s the right strategy.
How to Balance Depth vs. Breadth#
Should I become really good at one thing, or stay generalist? I’m leaning toward going deeper in operations research and healthcare applications, but part of me worries about missing out on other opportunities.
How to Build Things That Actually Matter#
I can analyze data and build models, but I struggle with turning that into something people actually use. I want to learn more about product thinking, user experience, and how to build systems that solve real problems at scale.
For Other New Grads#
If you’re just starting out like me:
- Don’t stress too much about picking the “right” specialization—just start building things
- Use AI tools to learn faster, but make sure you understand the fundamentals
- Find real problems to work on, even if they’re messy and unglamorous
- Get comfortable being confused—I’m confused about new stuff constantly, and that’s apparently normal
- Stay curious and keep building—the field changes fast, but that also means there are always new opportunities
I’m still pretty early in this journey, and honestly, I have more questions than answers. But I’m excited about all the stuff I don’t know yet and eager to keep learning. The combination of mathematical thinking, real-world problem-solving, and rapidly evolving AI tools feels like we’re at the beginning of something really interesting.
The key is staying humble about what I don’t know while being proactive about learning it.
View My Projects