In our new blog series, Industry Leaders on the Future of Data Science we asked industry leaders from across the UK what they expect to see from data science in 2022.
In this edition, Davide Anastasia, Head of Data at Audigent, shares his thoughts on how businesses and data scientists can prepare for the coming year.
As Head of Data, Davide looks at advertising ecosystems and tackles problems including consumer identity and advertising targeting against the backdrop of tighter privacy laws and changes around how all major web browsers handle cookies.
Davide looks at how Audigent can build innovative solutions that put data privacy and individual data rights in the spotlight.
Davide, thanks for joining us. Looking back on 2021, what do you think the biggest challenges for data scientists are right now?
With the widespread usage of Auto ML/AI products, we might experience a false sense of security. A few years ago, many thought that using NoSQL or Hadoop would have magically solved their scalability and data volume issues, and how many sad stories have we heard of failed projects?
In my opinion, you can use a technology when you have studied its internals and you understand
Data science is now going through the same phase. I personally believe it will be quite some time before the experience that is currently know-how of a few will be widespread enough and common knowledge across the entire industry. The improvement of libraries (especially usability) and tooling will go hand in hand with this process (we already see this with products like feature stores or experiment tracking).
There have been a lot of changes in the past 18 months, which ones do you think are here to stay?
The office as we remember it won’t be the same ever again. Companies that will force people to return to the office will lose their best employees (look at what is happening at Apple!) in favour of companies with a more agile mindset and a more modern infrastructure that allow people to seamlessly choose where, when and how to work their hours.
For individuals like myself, that follow both technical and people development, the way we structure projects and evaluate performance will also change massively. Overall, though, I welcome the change, as I think this will give more people a chance to shine, but also the opportunity for everyone to find the right pace to give their best.
We’re at the end of a year that’s seen remote work normalised and massive hiring of data scientists, what do you think the future holds?
A few years ago, cloud providers triggered the migration of software platforms from on-premise hardware to cloud resources. More recently, cloud vendors have pushed this boundary even further with serverless computing. I see the data science industry slowly picking up on the same trend, moving from models being developed in-house to training performed on the cloud. In the last couple of years many vendors have started offering Auto ML/AI features, essentially serverless platforms capable of “understanding” your data and offering the best solution for your problem, and I can see these kinds of platforms becoming more and more widespread, due to the very low barrier to entry.
How can data scientists and businesses prepare themselves for this future?
Be receptive to changes. Individuals and organisations have a different role in this equation: individuals have to push for their own development, starting from a strong theoretical foundation (and this applies to data science as any other skillset), while organisations need to take care of infrastructure, tooling and resources. When the two meet in a synergetic working relationship, individuals will propose ways to improve the organisation (for instance, current technologies to replace legacy ones), while organisations will provide the framework for individuals to expand their knowledge, either with the right set of projects and deliverables, or through the use of external training.
A lot of this is happening against the backdrop of remote working. What’s your advice for people leading remote teams?
It was true before we went remote, it is even more true now.
What makes a person tick is different for each individual; the way we communicate is different for each individual; the way we voice unhappiness is different for each individual.
As people managers, we have the duty to pick up these signals, understand the weakness and strengths of the people we have decided to surround ourselves with and find the right way to put the puzzle together, making sure that both team and individuals are happy and hence perform at their best.
In an office environment there are many ways to pick up on these things, but in a remote environment it is harder, so I have often found myself in the situation to ask very direct straightforward questions to measure the pulse: if you have built from the beginning a good relationship with your team, they will be willing to tell you honestly what is wrong and discuss a solution.
Once people are happy with the way they are spending a large chunk of their day (and there is more to a job than just your duties), any technical challenge is a secondary issue that can be easily discussed and solved.
What’s your advice for new data scientists breaking into the industry?
Essentially the same as any skillset: study, learn as much as you can from the others (wins and losses) and get ready for the next challenge. As always, try to understand why things are the way they are, because learning concepts is usually more fruitful than learning notions. In fact, concepts help you abstract problems, find commonalities, and generally help your brain to find patterns, while notions require just memory and can be very dogmatic (“we do this because this is what I have always done”.... how many times have we heard it?).
For this reason, I believe it’s more valuable in the longer term to study the mathematical foundation of the algorithms we use, instead of the specific implementation from a specific library.
Diversity and Inclusion is finally taking its place in the centre of many conversations. How can we improve it in the data industry?
I don’t think data science is necessarily different from other industries: we need to figure out where the barriers are and how we can bring them down, essentially tackling two areas.
First, we need to solve how we can improve access from the bottom, pushing (and help/fund) kids of any background to pursue a STEM curriculum. More broadly, we need to work on a system that helps people to find their vocation, and then pursue it. We need to give kids the possibility to experience the largest possible array of skills and we need to tell them there is nothing they cannot do. Once they find their path, being successful at what you love is easy (well, easier).
For us a bit “older”, we need to make the system fairer for those who are struggling now and help them express their potential, but also get ready for the arrival of the next generation, with the hope they can focus on solving important problems and make this planet a better place, instead of fighting for their rights.