Six Months as an ML Engineer

Last week I hit my six month milestone as a professional Machine Learning Engineer. Wow, even writing it, it doesn't feel real. I simultaneously feel like I started yesterday and that I've been doing this a lifetime. Even 2 years ago, I would never have imagined that this would be the career I'm in. I sure didn't plan to be here! I think I've finally understood why when I ask people how they chose their career they say "Oh, I sort of fell into it" or those motivational talks all have that image of the straight line and super curly messy one and tell you that your path to your destination is the latter and not the former.

I didn't plan to be here.

I might have got to where I am with some serendipity, but I do think that I'm settling where I fit it, without having to force it. Don't get me wrong, it's tough. I work hard to make progress and I am learning every day. It's not easy, just because it fits, but I'm starting to realise that's okay and isn't a poor reflection on me. It's just part of life and part of stretching yourself and a new field.

My favourite part of research was the analysis, figuring out what data I had and how I could make it useful; applying methods and algorithms to my data to get those results, drawing meaningful conclusions from them and coming up with and implementing the plan based on those conclusions. I wanted a job that would allow me to do all that i.e. one that didn't really exist.

At first I thought I wanted to be a Data Scientist, I thought that would allow me to do everything I wanted and, to be honest, there are probably some jobs out there as a Data Scientist that would. The thing I learned when I transitioned to industry (and am still learning sometimes!) is that you don't have to do it all, which was my experience through my research. I can just pick one part of the process to be really get good at, and still be a valuable member of a team with different skills to achieve that end goal. Actually, I love that about my job.

It turns out that actually the part I love the most is making something useful and facilitating the high standards and practices, which ties nicely to being focussed on MLOps. That's exactly what we are here for, and it's a lot of research too because as an industry we're figuring out what those standards and practices are still. I am still interested in writing models and the real-world relevance of them, and I like to know how the work I contribute to makes a difference. But, I don't think I need to be the one doing those things to appreciate them. My skills are not in business relations, supply chain management or statistics! I'm picky, I like things to be done right and I want to make sure that the "cool" and "smart" things people do are put to good use and actually add value. It sort of sounds like I fell into a speciality that was made for me.

Learning is a process.

The learning journey has been really new to be, you'd think after a 27 years of education and academia I'd know how to learn? Well, yes but it was different. In all my taught learning there was structure and it was my "job" full time. Even during my PhD there was more guidance in some ways because you had a supervisor and whilst I was there to research the learning was part of that. Learning as you work is a new thing to me. Actually, I should be seeing it similarly to my PhD - I'm here to do a job but part of that job is upskilling, especially in the tech world as things change quickly, and even more especially in ML and MLOps because things are establishing even quicker because it's relatively new.

Over the last six months I've had days where I feel like I'm floundering, but I'm pleased to say those are getting fewer and the days I feel like I'm achieving are growing. Both put the fear through me, even though, really, neither should. If I feel like I'm floundering, it means I've identified an area to focus my upskilling on - this is helpful guidance in such a broad field with so much to learn! If I feel like I'm achieving, it's because I'm implementing what I'm learned. Both are necessary parts of the journey.

How do I learn?

Firstly, I have a cracking team. Really, I honestly wouldn't be able to have learned even half what I have without them. Not only do we share learning resources amongst us for all our learning, but we have an environment where asking questions is encouraged. I feel safe to ask sometimes the simplest questions because I need to clarify something to help me understand the bigger picture. My team will send helpful articles in our chat, as well as have open discussions about things in wider meetings and knowledge sharing sessions. I find this has helped me learn things I didn't even think about learning or know how to learn them. That's what experience can give than documentation can't always.

I do some learning of my own too, Google and I spent a lot of time together. I like to find blogs, videos and articles to give me an idea of topics of interest/need. I like to find more human ways to absorb the information rather than going directly to any official documentation (if there even is any!) Of course, I do use the official documentation when I'm looking at the specifics but I always feel like I need a good foundation before I tackle the docs unless there is good "official" training material. For instance, whilst I use Microsoft Azure primarily in my role and use Microsoft Learn, I find the AWS and GCP training materials too for things that are platform agnostic (e.g. technical writing training!)

The community is also great to learn from, this might take a little bit of searching/travelling depending on where you live. I'm fortunate enough to have various meet ups that are organised that I can go to. There's a general tech meet up which helps to broaden my knowledge of the tech space generally, but also there is an MLOps community that meets every couple of months. This is my favourite one, we are a small group because there aren't many of us around but I find the talks/presentations really helpful for my learning. Chatting to others is also great, I'm not a big fan of networking to put it lightly, but learning what roles people are in and how MLOps is relevant to them helps me to understand further applications and see how it might develop in the future.

Key Takeaway

Even six months in, the fear goes right through me when anyone asks my opinion or advise like I might actually know what I'm talking about. I'm grateful that people think I am competent and have gained enough experience/knowledge to even be worth of asking advice from. It's a learning opportunity, and I try to take it as such and I am honest about where I'm not sure - worst case, I say I don't know and I go and look it up.