I recently met a brilliant data scientist who had built some amazing AI automation tools. But he kept saying things like “it’s not very original” and “I’m not building anything new, I’m just piecing together what’s there.” And I told him: “Your solutions don’t have to be original. They just have to solve the problem.” This is something I see all the time with technical people. They think they need to invent something revolutionary to be valuable. Here’s the thing… Your stakeholders don’t care...
9 days ago • 1 min read
Like most data scientists, I started my career in traditional 9-to-5 roles. With bills to pay and limited experience, a regular job felt "safe". Then COVID hit. Nothing felt "safe" anymore, and my dreams of working for myself no longer seemed so insane. I finally made the leap - and never looked back. Many data scientists share similar dreams of breaking free from traditional employment, but don't know where to begin. Yet, here's what I've learned along the way: there's no single path to data...
13 days ago • 1 min read
A data scientist I know once told me of a manager who would end every team meeting with the same advice: “keep kicking goals.” What does that even mean? Which goals? With what resources? How do you even measure what a goal is? Unless you’re actually talking to a football team, “keep kicking goals” is meaningless. Yet this is what some data scientists face. Well-meaning managers who say “go do data science” without giving direction on what that actually requires. And then they wonder why...
16 days ago • 1 min read
Want to know the book that's had the greatest influence on my thinking as a data scientist? I bet you'll never guess. It's Jurassic Park (the book, that is - definitely NOT the movie). As a teenager, I read everything I could find by Michael Crichton and his message of focusing on solving the problem while ignoring all irrelevant distractions still influences the way I work. Whenever I feel overwhelmed in my work, that's what I remind myself. I've been doing it for decades and will do it for...
20 days ago • 1 min read
There are three ways to fail as a data scientist. Only one is about technical skills. Solving the wrong problems; Producing the wrong results; and Not being recognised for the value your solutions create. If you’re not solving the right problems, then you have a opportunity identification issue. If you’re solving the right problems, but producing the wrong results, then you have a competency issue. If you’re solving the right problems and producing correct results, but nobody knows, then you...
23 days ago • 1 min read
"Genevieve, I need you to print out every page of this Excel spreadsheet by lunchtime." That was an "urgent" request I received from a senior leader during my graduate year as a data professional. I had been so excited - finally, someone needed my expertise for something important! This was my moment to shine and show my value. Instead, I spent the morning at the printer, watching page after page of spreadsheet data roll out, wondering how my expensive degree had led me to this. When I first...
27 days ago • 1 min read
Last week, I cringingly looked back at my old email posts from 2022, as I prepared to migrate to my new email list provider. The experience has been eye-opening. What struck me most was that they all seemed so… random. For example: What Robodebt Teaches Us About Responsible AI Business Lessons From “Succession” How to Avoid the Robopocalypse Design Principles for Data Science (from the Victorian Government’s Digital Strategy) It wasn’t terrible content, by any means. But unless someone...
about 1 month ago • 1 min read
Doctors were using AI-powered EKG software to confirm what they already believed, not to discover new insights. This was data scientist Dr Russell Walker's eye-opening discovery when working in the medical equipment industry. Physicians didn't want the sophisticated diagnostic capabilities his team had built - they just wanted validation of their existing interpretations. And if they didn't get that confirmation? They'd send the patient to a cardiologist anyway. That's the scary part - even...
about 1 month ago • 1 min read
5 years ago, my data science “superpower” was code optimisation. I could take a piece of analysis code that took 30 minutes to run, and with a few tweaks, make it run in 2 seconds or less. Spoiler alert: the secret is vectorisation using NumPy. It was a trick I picked up while doing my Computer Science Masters, where several assignments included strict runtime constraints. In fact, I got so good at doing this, that I often found myself spending hours optimising code, even when the potential...
about 1 month ago • 1 min read