A Day in the Life of a Pendulum ML Engineer

Life as an engineer can take you in many different directions. Where you choose to work defines what that can look like. At Pendulum, we’ve always embraced non-linear growth. We pride ourselves on equipping our employees with the mentorship, opportunities, and resources they need to pursue their true career ambitions. 

‍In this second installment of our Life at Pendulum series, Vincent Njonge shares his growth journey from his time at the company. Vincent first started as a Junior Data Analyst. During his time at Pendulum, his machine learning skillset has grown dramatically. Today, he works as an ML Engineer alongside our Data Engineers, Product Managers, and ML Scientists to develop and deploy the models that power Pendulum’s solutions.

Your growth path at Pendulum has been quite unique. Can you tell us how it started? 

I originally joined Pendulum in March 2022 as a Junior Data Analyst, having recently graduated from university. When I first started, my role mainly involved handling larger datasets and helping to build databases for our machine learning team.

Throughout the interview process though, I made it clear that I was keen to pursue a career in machine learning, and the engineering team took that into account early on. Within my first three months, I found myself working with increasingly complex projects and data sources. It was clear that the team was actively curious about where my skills were best suited. They soon tasked me with building a forecasting package that would allow us simplify our internal ML codebase and make it more reusable across our products. That was certainly challenging but proved to be a really important learning opportunity.  

Were you always interested in machine learning? 

Although I never actually studied machine learning as part of my degree, I’ve always been interested in the field. Before joining Pendulum, I had devoted a lot of time to independently studying it and participating in ML hackathons – I even won a few. 

This is something the team was quick to recognize. I spoke with our Director of Machine Learning during our company-wide Dev Week. I soon found myself becoming much more directly involved with the ML team – where I now sit today. 

What does your role now look like?

What machine learning engineers actually do is different wherever you go. At Pendulum, the role sees me take full ownership of some of our ML models right through to their end deployment for our customers. Here, we utilize an incredibly broad spectrum of ML capabilities – and that gives us a lot of scope to grow in the areas we’re most curious about. 

“We utilize an incredibly broad spectrum of ML capabilities, which gives us a lot of scope to grow in the areas we’re most curious about.”

Personally, I’ve always found time-series forecasting really interesting. Its real-world application is absolutely massive. That’s where I focus the majority of my attention at the moment and also where I see myself specializing in the long term. 

In what ways has your skillset grown?

Without an academic background in ML, I'd credit Pendulum with driving a lot of what I’ve achieved since joining. It’s the first company I’ve worked for, and in many ways, it feels like a natural continuation of my education. The culture here is based on learning new things, and I feel like I’ve been set up for success in my own personal development. 

“The culture here is based on learning new things, and I feel like I’ve been set up for success in my own personal development.”

I’m not just writing better code. I’ve taken really specific, in-depth courses in things like time-series forecasting to grow in the areas I’m most interested in. I’ve also learned a lot about collaboration and communication – which you also don’t always get in engineering roles elsewhere. 

What kind of customers do you work with now? 

I currently work in our Predict and Plan\Products pod – these are inventory forecasting and optimization solutions for commercial companies, but we also deploy them in the global health sector.

That’s a nice place to be. What’s good about it is you actually see the models we are building and deploying actually make a tangible difference. It’s not just about writing production-ready code, every prediction is actually improving some really important operations. 

What would you say to anyone thinking of pursuing a career in ML? 

Machine learning is an immensely broad field, so look for a place that will allow you to really grow in your skillset. There are still so many new frameworks and modules I can upskill in and learn from, and this is a place where I have space to do that. PyTorch, for example, is one I’d like to focus on next. With more experience in that, I’ll be able to review new forms of code and work more closely with our ML Scientists. Your managers and leads recognize that carving out time to develop in these areas benefits everyone, and that breeds a culture where you never stop learning. 

Can you walk us through what a day looks like for you? 

9:30 AM 

I’m based in Nairobi, where we have a coworking space, so I go in semi-regularly; today, however, it's a Friday, so I typically work from home. I start the day by catching up on emails and internal threads and checking the status of ongoing projects. There are a few action items on my side that I can amend quite quickly. 

11:30 AM 

I also have a few pending code reviews. I like to prioritize these in the morning so my colleagues on the same pod aren’t blocked on anything from my side. 

12:00 PM

Once those are completed, I’ll dive into my main technical task for the day. Because of our geographical location, the mornings are usually free of meetings, so it’s the best time to focus on this. Yesterday, I spoke with a colleague in the US to get feedback on an investigation task in ClearML. I’m now almost done with this and soon begin adding updates to the weekly pipeline for one of our customers. 

1:30 PM

I prefer a slightly later lunch – so I take a little bit of time out after I’ve completed my updates to get something to eat. 

2:30 PM 

I return in time for my stand-up with the rest of the Predict and Plan\Products pod. We each give updates on what we’ve been working on for the past few days or so. These meetings are also a great opportunity to discuss where we can help assist each other’s workflows. For example, a recent roadblock came up, so a colleague and I set up time to discuss it when we’re both in the office early next week. 

3:00 PM

Once I finish the last of the updates, I run one of our weekly forecasting models. Although most of these are automated, this specific one is still work-in-progress, so I trigger a manual run. While that’s happening, I’ll look ahead to next week, and prepare my ‘to-do’ list for the tasks that I’ll start on Monday.

5:30 PM 

The model successfully runs, and the completed predictions are sent directly to the customer. I update our JIRA board with a few additional items to address next week before logging off for the weekend. 

‍

While every day as an engineer can look very different, Vincent’s story demonstrates how Pendulum can accelerate your growth in the areas you are most passionate about. Visit our careers page to learn more about our machine learning team or to apply for a role. If you want to read more about life as a Data Engineer, read Jean’s story from our Life@Pendulum series. 

‍