Often described as a “pragmatic disruptor”, Dina Mohammad-Laity, an independent data strategy advisor and former Director of Data Science at Talabat, says data is an inherently creative industry – creative problem solving, creative storytelling.
Interestingly, she believes in de-hyping all parts of the data lifecycle. In this freewheeling interview, the Dubai-based data expert discusses the increasing representation of women in data engineering and analytics, keeping ethics and integrity at the heart of data projects and the challenges in finding right data talent in the Middle East.
How do you see data science impacting and changing the world we live in today? Can it be misused?
I see data-driven impact and change everywhere – it’s ubiquitous, and not just in the workplace. Think about how much monitoring and advice you take from your wearables, like a fitness watch, or how much of the cognitive load of choosing a movie is taken by Netflix’s recommendation system. Data Science is becoming fully integrated in our lives.
In a professional setting, applying data to solve problems is no longer a “nice to have” but a requirement, and the increasing representation of data professionals in leadership roles is a testament to how the business landscape is changing in that regard.
Of course, with great power comes great responsibility, and misuse of data science is, unfortunately, common. It’s important to keep ethics and integrity at the heart of data projects. For example, improper profiling can lead to further discrimination against already marginalised groups, even unintentionally, and personal data storage can bring huge security risks. Businesses must invest in ethical frameworks and security to ensure that their goals are not achieved at any cost.
In 2020 it was reported that women fill up about 15- 22 per cent of data scientist roles. Do you feel efforts to improve female representation in this industry are starting to build momentum?
I’ll have to answer with a known level of selection and recency bias here. Over the past year, I’ve encountered a rapidly increasing proportion of female data professionals. Not just in the data science space, but related roles like data engineering and analytics. That’s encouraging to me because the data industry offers a vast array of stimulating, impactful and well-paid jobs that suit many different skill and attitude types. I always like to think of data as an inherently creative industry – creative problem solving, creative storytelling. So the fact that we’re undoubtedly seeing more women come into the field means we’ll all benefit from new types of creativity being the norm.
Generalists or specialists: who fulfils the business demands best?
I don’t think it’s either/or — data is a big tent! It depends on the stage and needs of the specific company. I once saw a tweet that said: “build your startup with generalists. Scale your startup with specialists” which I can see applying well across data hiring generally, although even more applicable is the idea of “don’t do data science, solve business problems”.
More than generalist vs specialist, I like to consider pragmatist vs idealist – in most cases the pragmatist will win out, as you need talent that can work with the world as it is, not as it should be.
When it comes to recruitment, I like to look for data professionals that are “T shaped”; people who have a breadth of skills and knowledge, with depth in one or two areas. These are the people that I find work well in cross-functional teams, and often have a curiosity and “figure-it-out” mindset that is so valuable in the data industry today.
Do you also feel that the right kind of talent is a challenge in the industry?
It depends on the level. It seems like a lot of really great junior talent is coming into the job market, which is awesome to see, and many of these individuals have that “T shape” already, even early in their career. The struggle is more with senior roles – the supply just cannot keep up with demand, so it’s worth being creative with your talent acquisition.
Investing in existing team members to bring up their skill levels to the business need may fill some gaps although this can take time. When it comes to hiring, it’s worth taking a step back to make sure you’re recruiting for the right type of person as there are greater scarcities in some talent pools vs others, and some skill sets are easier to train on the job. To paraphrase Cassie Kozyerkov, Chief Decision Scientist at Google, “Are you in the business of baking bread or building ovens?”
In the Middle East, is data management still a fundamental challenge, even though enterprises are becoming data-driven. Why?
It certainly relates to the earlier question on finding the right talent — skill shortages are across the board. I also find there’s often a phasing issue — who to hire, in what order? I’ve seen companies end up with a number of talented data scientists, but no data engineers or data platform to work from. It’s backwards. You need to start with clear and structured frameworks to deliver data solutions that are tied to value, secure and reliable.
The movements in the “modern data stack” approach have helped in this space a lot – you can get going much faster, with lower staffing requirements, as today’s tooling is better suited to businesses’ needs.
What are the most popular tools and platforms used in the AI and analytics industry in the Middle East?
I can’t speak for the entire region, but I can share what I’ve been working with a lot lately. In terms of data platform development, I’ve been using Stitch Data by Talend for ingesting various sources in data lakes – this simplifies pipelines a lot, and reduces the need for many handcrafted extract and load jobs, which improves reliability and reduces manual intervention.
Snowflake and BigQuery have been consistent themes over the past couple of years for cloud-based data warehousing. My “tool of the year” for 2021 has to be dbt — it’s a tool that allows transformation within the data warehouse itself. They recently announced an integration with continual.ai which aims to bridge the gap between ML and business analytics. I’m excited about the impact this can have across a number of industries.
What’s the most exciting challenge, or set of challenges, that you see on the horizon in data management? What are the possible solutions?
I briefly touched on it in the question about tooling, but it’s the era of the “Modern Data Stack” that is exciting me, because it’s solving a whole range of challenges that have persisted over a number of years. Accessibility is a big challenge in my eyes – we need to get more people using data to make smarter decisions, but how can we do that if the industry is perceived as impenetrable, or guarded by gatekeepers? By teaching people how to climb fences! Some of the new tooling coming into the market is bringing access to sophisticated data methods, and high impact outcomes, to the mass market. That can only be good for us all as better decisions are made more broadly, with data being less of a bottleneck.
How is big data evolving today in the industry as a whole? What are the two most important trends that you see emerging in 2022?
The data space is evolving to be a core part of how we live, and how we do business. I think in 2022 we’ll see an ongoing rise in predictive analytics, especially as the tools to deliver this effectively become commoditised and easier to roll out (see BigQueryML for a good case in point). I also think we’re going to see increased investment in data orgs, in tooling and talent, as the impact of these investments is felt and measured across all parts of an organisation.
What data science and analytics podcasts do you listen to?
I am a podcast and audiobook fiend — I’m constantly listening to something. Some of my favourites are:
- ThePresentBeyondMeasureShow-LeaPicadoesagreatjobwith her guests talking about data storytelling as a field in its own right
- DataSkepticcoversawiderangeofapplieddatatopicsinanengaging and accessible way
- NotSoStandardDeviationswiththelegendaryRogerPendandHilary Parker keeps us up to date with trends in the data industry as a whole (so you don’t have to spend all your time on data twitter to understand the in-jokes!)
If you liked reading this, you might like our other stories
What Marketers Don’t Know About AI
Security Is A Big Data Problem
