A Journey in Scaling AI
Gabriel joined Ocado Technology in 2020 as Chief Data Officer, bringing over 10 years of experience in leading data science teams and helping organizations realize the value of their data. At Ocado Technology his role is to help the organization take advantage of data and machine learning so that we can best serve our retail partners and their customers.
Gabriel is a guest lecturer at London Business School and an Honorary Senior Research Associate at UCL. He has also advised start-ups and VCs on data and machine learning strategies. Before joining Ocado, Gabriel was previously Head of Data Science at the BBC, Data Director at notonthehighstreet.com, and Head of Data Science at Tesco.
Gabriel has a MA in Mathematics from Cambridge and an MBA from London Business School.
At the moment Demetrios is immersing himself in Machine Learning by interviewing experts from around the world in the weekly MLOps.community meetups. Demetrios is constantly learning and engaging in new activities to get uncomfortable and learn from his mistakes. He tries to bring creativity into every aspect of his life, whether that be analyzing the best paths forward, overcoming obstacles, or building lego houses with his daughter.
Gabriel talks to us about the difficulties of scaling ML products across an organization. He speaks about differences in profiles of data consumers and data producers, and the challenges of educating engineers so they have greater insights into the effects that their changes to the system may have.
There’s a time and place for everything. Seasoned ML engineers like Gabriel Straub may have just mastered that time and place for machine learning. As the current Chief Data Officer for Ocado, he brings over 10 years of experience leading data science teams and helping organizations get the most of their data.
We highly recommend readers and viewers (and listeners when they get home) to watch this quick introduction video to what Ocado Technology does. It really changes the way you view this interview when you see 1000s of robots running simultaneously!
This perspective requires a mature machine learning practitioner because noobs like me want to implement machine learning in everything. A seasoned developer knows that there’s a time and place for everything.
Simple Rule-Based Architecture is a Good Baseline for Machine Learning, So Start There First
Aspiring to implement ML effectively, often means starting simple. It’s hard to implement and having a strong foundation can make your life so much easier later. Gabriel doesn’t think you shouldn’t use machine learning for everything, but that you should start simple. Start with rule-based architecture that’s open for machine learning updates later on. By doing this, you now have a baseline metric (rule-based system) to compare the results of your new ML update in the future.
Starting simple may not be as cool or resume-padding, but it serves as a good foundation.
Ask the Right Questions
You’ll never get the answers you need if you don’t ask the right questions. Gabriel seems to have a knack for asking the right questions. What’s the cost of getting it wrong? How do you give the right feedback loop to producers?
What's the cost of getting it wrong?
It’s a valuable question for figuring out your priorities. It also helps you reconcile how to wisely utilize the resources you have. It forces you to focus on the task at hand, evaluating your ability to actually solve a real-world problem, rather than using the newest technology. Sometimes the old-school way works just fine.
How do you give the right feedback loop to producers?
Communication between teams and stakeholders is always a hot topic. Asking yourself this question will probably save you a lot of headaches later as you’ll already have everyone on the same page when the going gets tough.
Make Business Users Feel Empowered When Building Applications with Machine Learning
A successful machine learning project depends on the value you bring to the user/business, not specific ML metrics.
It’s not enough to build effective ML systems if you aren’t hyping the value-add it provides. Gabriel tells two stories about sharing machine learning tools with business units and whether or not they adopted the tool. For the first encounter, his team shared the tool and told them how smart it was and how complex the machine learning within it is. These users felt intimidated and as though they were being replaced by a tool.
For the other encounter, his team focused on how this tool would help the business users do their job better and faster - relieving a lot of manual labor and headache. Although this was the same tool, Gabriel’s team did not focus on the machine learning part, but on the value-added part. They explained how the tool would make their lives easier. The second business unit adopted the tool!
Building an amazing machine learning application is worthless if no one uses it and it probably doesn’t pad your resume the way you think it does!