Data Scientists are in demand. And while that job title is applied to all sorts of disciplines, the reality is that nearly anybody with some analytical and predictive capabilities can (safely) call themselves a data scientist.
In this post, I’m going to focus less on what a data scientist is (lord knows there are enough posts about that) and more on what it means to be a freelance data scientist, what I do, and why I choose to do it.
Insight for hire
A lot of smart people think that data scientists should have domain expertise. That they should be deeply familiar with the problem area that they’re working on. I agree with them.
The best person for any analytics project is someone with two things:
- Deep domain expertise
- Analytical / statistical knowhow
I’m just going to come out and say it - I am not an expert at anything. And yet this doesn’t rule me out from being of help to those who hire me.
The reason for this is I’m not really a freelance data scientist. That word (freelance) simultaneously tells you everything you need to know about how I work whilst also saying precious little about what I bring to the table.
A better term would be somewhere in between consultant and facilitator. I bring technical expertise, knowledge of what is attainable, and a general can-do attitude to help groups of actual experts make better decisions.
I combine my skills with theirs. And when necessary, I bring any additional talent needed to get the job done and align them with the project, too.
This means that there isn’t a system, or a consulting framework, or a specific deliverable that I provide to my clients. This means that no two engagements are alike. This means that what I actually do changes very often.
What I do
It’s been 10 years and I’m still not sure exactly what it is I do. It seems to me that the longer you do something, the more abstract your capabilities become. Every so often I cave and look for full-time, permanent jobs. They want X years of Python, Y years of scikit-learn.
The further I get into this journey, the less I think in terms of frameworks, languages, or packages. What I say today is ‘I help businesses discover the value of advanced analytics’ and that’s been true for basically my whole career.
You may think that this sound wishy-washy - a throwaway cliche. It’s not. That really is where I add my value.
When major enterprises hire CTOs or CIOs, they aren’t looking for a certain number of years’ technical experience in ASP.NET or Java. They’re looking for strategic help, management skills, pragmatism, and a whole host of indefinables.
To anyone who’s reading this and considering becoming a freelance data scientist - please try to develop the indefinables. Technical skills are a dime-a-dozen. Yes, even in data science. A quick search on Upwork will return a long list of people more experienced than you, some of whom who are willing to do the work for cheaper.
You can’t compete with these people in that way. The good news is that I’ve found a lot of these people to be one-trick ponies. They have to be to charge so little. They’ve narrowed so far in on Tensorflow, or churn prediction, or credit scoring, or chatbots, that they can’t transfer their skills to different problems.
Problem solving will never go out of fashion.
In a decade of freelance analytics work I’ve built data warehouses, reports and dashboards, spreadsheets(!), computer vision applications, recommendation engines, chatbots, server-side software, and SaaS applications. I’ve compiled and written documentation, I’ve managed teams and projects. I’ve done a lot of different tasks.
Very often, I’ll pause and consider where I am in my career. I’ll make a plan to do more of this and less of that. I’ll draw out a 5 or 10 year plan. Then a client will call, we’ll discuss something exciting, and I’ll throw my plan out of the window.
My career so far has been one of doing the coolest, and most impactful, things I could do at that moment. I hope that continues for a long time.
How I find clients
I don’t want this section to be an eventual disappointment, so I’ll make it an upfront one - I don’t know how I find clients.
I’ve never had the patience for creating AdWords campaigns, building content marketing strategies. I’m hopeless at networking. I despise Upwork.
Over the past 10 years, I’ve been out of work a fair amount. I think pretty much every freelancer that’s been working for themselves that long as had some dry patches. There have been some lean times and some very, very busy periods. A lot of people can’t stomach that. They cut their rates, they develop involved sales strategies, they ‘reach out’ to everyone they know on LinkedIn. I wait. I write and I work and I wait.
For 10 years, I’ve eventually found work by staying intrigued, having interesting conversations, and waiting. If you don’t wait (or can’t wait), you’ll take work that doesn’t fulfil you and you’ll make ends meet (maybe). Do enough of that though, and you’ll eventually meet your end as a freelancer.
Choose is an active verb
I didn’t call this post ‘Why I chose to be a Freelance Data Scientist’, that was on purpose. The thing about being a freelance anything is that it takes near-constant effort. I know there are many people who put a great deal of themselves into their work, but there are also those that cruise through their work weeks.
Every time I speak to client, send a proposal, or write a blog post, I’m committing again. I’m choosing to be a freelance data scientist.
If work dries up, it’ll only take a day or two of Netflix binges before I get bored and figure out a new problem to solve. As long as there are difficult data problems to solve, I’ll choose to be a freelance data scientist.