Q&A with Machine Learning Engineering Lead Mouloud
In our first installment of Day in the Life, our series exploring the life of our Data and Analytics (aka DnA) team at Vistaprint, we meet Mouloud Lounaci, the Machine Learning Engineering Lead in the Pricing, Promotion & Personalization Domain.
Mouloud is a machine learning enthusiast who loves designing data products to make the life of Vistaprint customers easier. Working from home in Barcelona, he implements cloud-based data solutions and deploys ML-powered applications in a data mesh organization.
Mouloud – you’ve been working as a data engineer for seven years now. What led you to data engineering and, more specifically, into machine learning?
It’s a funny story. I kind of got into it by accident. I went to school to study software engineering and then in the last year of college, I took a class in machine learning and big data. And I thought, “Wow, this looks awesome. I want to work in this.” I saw the potential impact and use cases of machine learning that have simply not been seen before.
Even today, I think you can see a lot of things that we can do with machine learning. Like automatic translation or computer vision. Just a few years ago, we didn’t even know it was possible. What I love about it is that you’re working on something that didn’t exist and then making it a reality.
That sounds very futuristic. What state-of-the-art technology use cases can we look forward to seeing?
Today, we’re working with transformers for natural language processing (NLP). It’s basically the ability to understand the semantics behind text. In the past, we were processing text as words, without understanding its meaning or what’s behind those words. Today we have the technology to embed meaning into a machine and ultimately, make a connection between different texts.
A typical use case of transformers-based NLP for an eCommerce website like Vistaprint is what we call ‘Natural Search’ or natural language based search–as opposed to keyword based search. If you type in a complex sentence in the search bar, it can help us get you results that match your search better because the machine has a deeper understanding of what you’re looking for. And that just makes the customer’s experience better.
I think you can see the impact every day-to-day. For example, a restaurant owner can describe what he needs, say “minimalist business cards for my vegan restaurant business.” And instead of having to select the business card product gallery and then filter on restaurant industry and minimalist style, we can show relevant design templates for vegan restaurants and minimalist styles on the relevant product. All thanks to transformers-based models.
Sounds like the conversation around machine learning is always evolving. What are some hot topics that data engineers are talking about right now?
One of the current hot topics around machine learning that I am interested in is MLOps. MLOps is the process of taking a machine learning model to production to solve real life problems. It considers what you have to build to actually have a machine learning model deployed at scale.
For example, to have the NLP use case that I mentioned before in production, we not only need to build the infrastructure to continuously process large amounts of texts to train a model that is able to understand semantics. We also need to build an API to be able to use the model on the website, while continuously monitoring the model performance so that it continues to become more accurate over time. It examines all the operations that you need to set in place to have a machine learning model working.
And how does the conversation around MLOps inform your work at Vistaprint?
MLOps is derived from the broader concept of DevOps, which is a set of tools, concepts and ways of working for software teams to reduce the time-to-production of new features. It aims to ultimately deliver faster increments of value to end users and get their feedback more often through experimentation.
Machine learning development today is still slow compared to software development, we can spend several sprints between model inception to first A/B test and several more sprints to test new versions for that same Model. MLOps tries to bridge that gap between software development and machine learning development.
In DnA, we are on a journey to become customer-centric where the Vistaprint customer is at the center of everything we build. And MLOps is an enabler for faster iteration, from inception to experimentation, to make sure we are building the right models for our customers.
Let’s talk more specifically about your role at Vistaprint. What led you to explore data engineering here?
What made me choose to join DnA is the ambition to make Vistaprint one of the world’s most iconic data-driven companies. We are on a journey of transformation that brings exciting technical challenges and amazing opportunities to build ‘kick-ass data products’, in the words of our Chief Data Officer, using the latest data tech stack.
Something else that inspired me to join is the bold decision to be part of the pioneers in the tech industry to becoming a remote-first organization. And more specific to DnA, revisiting the traditional monolithic data team to move to a data mesh organization.
I’ve also discovered how Vistaprint meets a variety of customer needs. Whether that’s a small business owner who needs print marketing such as eye-catching menus and trade show displays, a brand-new logo or an online presence. Vistaprint’s products and solutions can help these businesses thrive.
There’s so much going on in DnA behind the scenes to make this happen. Such as optimizing the manufacturing process through accurate forecasting and developing a personalized customer experience that make Vistaprint feel like an extension of our customers’ teams.
Yours is a global team – how do you collaborate and connect with colleagues across the world?
This is also something that I love about Vistaprint, we have team members around the globe, from Boston to Bangalore, which brings an awesome diversity of thought and perspective.
In the Data Engineering chapter, we have a meeting every Monday bringing together the community of data engineers to talk about what new technologies we should be looking at, what issues we should be solving in DnA that will benefit all data product teams, and just sharing project progress, celebrating wins or surfacing blockers.
Every other week, we also have what we call a “Show DE,” which is a typical “show me” presentation for data engineers. It’s where we present any projects we are working on or share learnings. It’s a chance for everybody to ask questions and learn something new.
And then, within the data product teams, we have the Agile ceremonies. Our team members are distributed between the U.S., Europe, and India. And maybe soon Australia. For example, 2pm (in Barcelona) is, typically, what works best for most teams for Daily Stand-ups. It’s the morning in the U.S., the beginning of the afternoon in Europe, and late afternoon in India. That’s when we can get together as a team and really discuss the larger ideas on how we’re progressing on certain projects.
We also have virtual coffee breaks twice a day and participate in regular team building activities, like organizing chess tournaments, because one of our team values is “having fun at work”. All those things really help us stay on the same page and feel connected.
Finally, something I am looking forward to is to be able to meet in person again both for work in the collaboration centers and of course for awesome team buildings.
Finally, what would you say to someone who may be excited about Vistaprint’s ambition but not sure about working from home full-time?
Honestly, before joining, I wasn’t sure about remote work. I always thought that, at home, there are too many distractions, and it would be hard to be as productive as in an office.
What I’ve learned since I joined is that, with the right home office space, it’s the exact opposite. I am much more flexible to organize my day around optimal focus time and meeting time which results in much more productive days.
There are still challenges for me to figure out like the switch between work time and family time or how to replace the “coffee machine chats” but it brings so much flexibility in my life and I am enjoying more quality time with my family that it will be hard for me to go back to an office-based work.
Also, as Vistaprint’s collaboration centers start to reopen in Barcelona and other locations, we can have the best of both worlds—remote work for productivity and flexibility, and collaboration centers for in-person conversations and relationship building.
Are you a Data Engineer looking to join Mouloud and the kickass team in DnA’s Data Engineering Chapter? Be sure to check out the latest Data Engineering roles in DnA.