This October I complete ten years in Data Science.
I started as an Electrical Engineering graduate – no formal Computer Science background, no structured ML curriculum, no ChatGPT to ask at 2AM when nothing made sense. Just a zeal to establish that for my tribe, life exists outside the Power Plant and Power cables as well. There are also transformers which ignite people’s minds in the darkest hour. And the belief that I could figure it out, and an unreasonable number of late nights proving it.
Electricity flows. So does money. Both are invisible. Both power real lives. And when either gets disrupted, real people feel it. That’s what I protect now as part of my role as a Staff data scientist in fraud prevention.
What I want to write about is not the success – it’s the structure underneath it. What I actually learned. What the field really demanded. What stayed constant underneath all the noise.
This is that reflection.
The field has changed beyond recognition. On-prem computation to cloud. Classical ML to deep learning. Transformers to GenAI to Agents. Each wave brought new tools, new vocabulary, new things to feel behind on.
The list is endless, considering this is such an evolving field. But underneath all of it, I learned three life sessions which never moved.
Problem solving is an art, not a math exercise.
The biggest trap – and I’ve fallen into it – is arriving at a problem with a solution already in mind. Someone hands you a dataset and you start thinking models and dashboard. The real work hasn’t started yet.
Wrong problem, perfect model, zero value. The math is the easy part.
Curiosity is not about keeping up. It’s about being willing to be wrong.
Staying curious about new tools is easy. That’s just FOMO with a growth mindset label on it.
The harder version is staying curious enough – when you’re senior, when people are looking to you for answers – to admit “I might be completely wrong about this.” That kind of curiosity gets rarer the more you know. Which is exactly when it matters most.
Thus this space requires continues learning and un-learning, which is also a foundation of problem solving.
Believe in yourself — not the poster version.
The kind that keeps you going at day or night about something that isn’t working yet. When the fraud trend isn’t subsiding, the business/product owner is losing patience, and you’re not sure if the problem is the data, the approach, your AI agent or you.
Nobody talks about that stretch. But that’s where careers are actually built.
The next decade will look nothing like today. These beliefs reflect where my thinking stands today. As with any good data scientist, I’m always open to being proven wrong.
I am grateful to Adam Grant for introducing me to the importance of rethinking in Think Again.