Data Futurology - Leadership And Strategy in Artificial Intelligence, Machine Learning, Data Science

Organisations face multiple challenges when it comes to building teams in 2023. On the one hand, there is a skill shortage in just about every field of data science and analytics. Finding and attracting the best people to the organisation can be difficult.

On the other hand, there is mobility between jobs unlike anything we’ve seen before. The “Great Resignation” is still a major trend sweeping across Australia, and employees will be more than willing to move on if they don’t feel like they’re getting what they need from their jobs.

In this episode with Data Futurology podcast host Felipe Flores (a Chief Data, Analytics & Technology Officer himself), he explores both sides of this particular coin. In the first half, Felipe shares key insights and tips on how to recruit the best talent, including mistakes that he’s made in hiring and how he now looks at the interview and hiring process.

The second half of the podcast is dedicated to providing tips for retention. Contrary to the popular view, it’s not always a matter of remuneration. Indeed, studies consistently show that this is far less important to many employees than things such as the opportunity to build their skills or engage more deeply with their organisation. As Felipe says “People might want to become a product owner, or a strategic person that interfaces with the business and helps them to contextualise the results to the organisation.”

Tune into the podcast for these insights, and many more.

Thank you to our sponsor, Talent Insights Group!

Join us in Sydney for OpsWorld: https://www.datafuturology.com/opsworld

Join our Slack Community: https://join.slack.com/t/datafuturologycircle/shared_invite/zt-z19cq4eq-ET6O49o2uySgvQWjM6a5ng

What We Discussed:

00:00 Welcome to Data Futurology

0:29 What to expect from our upcoming event on operationalizing security for business value, impact and scale, at the Sofitel Wentworth in Sydney on March 14 and 15.

2:20 What makes hiring so challenging.

2:50 Three tips for hiring. Tip #1: Attitude.

3:40 Three tips for hiring. Tip #2: Transparency and openness.

5:53 Three tips for hiring. Tip #2: Be impressed with one technical thing in one technical area.

7:54 Why retention is important, and what is being done to improve it?

9:54 Three tips for retention. Tip #1: Provide formal training.

10:24 Three tips for retention. Tip #2: Give employees exposure to new work/projects.

11:31 Three tips for retention. Tip #3: Provide on-the-job training

Quotes:

This week on the Data Futurology podcast we are thrilled to welcome Tamara Mirkovic. Mirkovic is a specialist in data platforms, AI, machine learning and predictive analytics, and was recently the recipient of the Women In Digital National Award.

Mirkovic has experience building all-of-business data strategies and leading the change management program to get the organisation motivated behind the transformation.

It’s not easy, as Mirkovic said. When individuals and teams are already capable with data within their own silos, adopting a new platform can raise concerns for everything from cybersecurity to job stability. Even with the support of a visionary leadership team driving efforts from the top down, the success of a change management program relies on finding the right people to act as the champion within the peer group.

From there, it’s all about building a repeatable approach to the applications and models that will allow it to be rolled out to other teams across the organisation. “When you’ve got a critical mass of use cases, then everything going forward can be a version and iteration on what’s already been done,” Mirkovic said. “Our intention is to create enough patterns that can then be reused very quickly for other use cases.”

Finally, Mirkovic discussed what it means to have won the Woman In Digital award, and the challenges and opportunities that face women in the sector in general.

Tune in to hear these insights, and many (many) more from this wide-ranging and detailed interview!

Please note: The opinions expressed by Mirkovic in this podcast are hers alone, and not a representation of the organisations she has worked for and mentions.

Enjoy the show!

Thank you to our sponsor, Talent Insights Group!

Connect with Tamara:  https://www.linkedin.com/in/tamaramirkovic/

To learn more about Women in Digital: https://www.linkedin.com/in/tamaramirkovic/

https://womenindigital.org/

Join us in Sydney for OpsWorld: https://www.datafuturology.com/opsworld

Join our Slack Community: https://join.slack.com/t/datafuturologycircle/shared_invite/zt-z19cq4eq-ET6O49o2uySgvQWjM6a5ng 

What We Discussed:

00:00 Welcome to Data Futurology

2:36  Introduction to the podcast and guest, Tamara Mirkovic.

3:09 An overview of Tamara’s experience and current role.

4:08 On the subject of building a data strategy program from the ground-up: what has been the most effective approach?

5:16 When looking at a wide-scale transformation project, how do you break it down?

7:25 What are some of the major challenges that you’ll come across when undergoing digital transformation and data transformation?

11:42 What can be done to encourage the data scientists within the organisation to change the mindset around to use common tools and approaches, to build a more collaborative environment?

17:55 How did the leadership team support the initial idea, and then the resistance that came at first?

20:22 How did you go about finding that group of champions that would help drive the change management and success of the project?

23:34 What are some of the common fears that come up during change management, and how do you address those concerns? 

The past 18 months has been a period of unprecedented innovation across data science, machine learning, and AI. The depth of research and what has been brought to market has empowered data scientists in ways that, even a year ago, could not have been predicted.

Looking forward to the next 18 months, the industry is not going to rest on its laurels, but the question is where the next waves of innovation will come from. That is what Felipe discusses on this episode of the Data Futurology podcast.

He highlights four areas in particular where he would like to see the industry focus its innovation. Starting with the ease in which to undertake data preparation, and moving through to developing better machine learning ops and engineering, the combined “key areas of innovation” would allow people working in data science and beyond, into citizen data science, to better leverage the opportunities of AI and machine learning, and at speed.

Felipe then rounds the discussion out with a look into the ethics of data science. There is a lot more discussion that needs to happen in this area, he argues, so that organisations of all sizes across the world can be sure that they’re delivering the models that have the positive impact on the world that we’re all looking for.

It’s going to be an exciting year ahead for everyone involved in data science! Tune into the podcast for more insights.

Enjoy the show!

Thank you to our sponsor, Talent Insights Group!

Join us in Sydney for OpsWorld: https://www.datafuturology.com/opsworld

Join our Slack Community: https://join.slack.com/t/datafuturologycircle/shared_invite/zt-z19cq4eq-ET6O49o2uySgvQWjM6a5ng

What We Discussed:

00:00 Introduction

1:40 An overview of the innovation that has been brought to data science over the past few years.

2:33 Four areas where the industry can innovate further #1: Making it easier to do data preparation.

4:40 Four areas where the industry can innovate further #2: Democratising AI and empowering the citizen data scientist.

6:45 Four areas where the industry can innovate further #3: Automated machine learning can still be improved.

8:33 Four areas where the industry can innovate further #4: Better machine learning ops and engineering is important to being able to reliably deploy, monitor track and alert.

12:22 Do we have the data that we need to make the responsible models that we want to?

Quotes:

This week on the Data Futurology podcast we speak to Sanchit Juneja, the Director of Data Science and Machine Learning at Booking.com. Having worked in roles across SE Asia and Africa before landing in his current role in the Netherlands, Juneja has a truly world view of the role of data in business, especially within the context of product development within large enterprises.

One of the challenges that large enterprises face with product and data is the question of whether you should build or buy the tools that the organisation uses. As Juneja states, the ideal approach is a holistic one that does focus on speed to market. “You do still want to build things that are strategic and core to your heart,” he said. “However, having access to things that bring you faster to market is important at the end of the day, as you want to unlock business value.”

Juneja then shares insights around the skills that it takes to work in product management. Compared with some other areas of data science, it is perhaps not quite as important to be technical (though being “tech aware” is essential). However, those in product management need to be very good at building consensus across the organisation, from executive right through to those that will implement solutions. They also need to be very comfortable with ambiguity and working with the unknown and have an appetite to learn on their feet.

Finally, Juneja shares his insights around how he and his team track the value that they’re adding to the organisation. Critical in building alignment across the business is the ability to show results, so everyone working in product needs to be able to clearly articulate the gains there.

For these insights, and many more in the wide-ranging interview with Juneja, tune into the podcast!

Enjoy the show!

Thank you to our sponsor, Talent Insights Group!

Join us in Sydney for OpsWorld: https://www.datafuturology.com/opsworld

Join our Slack Community: https://join.slack.com/t/datafuturologycircle/shared_invite/zt-z19cq4eq-ET6O49o2uySgvQWjM6a5ng

What we discussed:

00:00 Introduction

03:31 What is the main focus of your role on booking.com?

07:42 How have you found getting the stakeholders and team members on board for the journey?

09:08 How do you define the product work in in our space?

11:57 What are some of the other skills that you see as key for the product manager role?

13:28 How do you make the trade-off decisions around the product as you're implementing or building towards the vision? What are what are some of the trade-offs that need to be done in the in the product decisions?

14:30 What is the mindset shift that that you would recommend for people that may be doing ad hoc pieces of work, or a one off?

16:07 What are you most proud of?

17:14 What are some things that you would have done differently?

Quotes:

· Even if you're a big tech org, you don't necessarily need to build everything yourself. So the build versus buy call is something that is personally on top of your mind, if you're a product leader. There are so many things that are happening, so many core things that if you go on and build it inside your house, it will take you the next six months. But if you just buy a tool outside in the industry, it will be much quicker for you. I think that is one thing that is always on top of your mind, what to build versus build to buy.

On the Data Futurology podcast this week we have AI expert and author, Cortnie Abercrombie. Abercrombie is the CEO of AI Truth, an organisation that empowers business leaders to leverage AI in an ethical and innovative manner. She is also the author of What You Don’t Know: AI’s Unseen Influence On Your Life And How To Take Back Control.

We start the conversation on the podcast talking about the challenges that data scientists face with data governance, and the many challenging questions that complicate that.

Then we discuss the challenge of maintaining models, and what that means for the safe shepherding of data. As Abercrombie notes, the average tenure of a data scientist at an organisation is only 12 to 18 months. When an organisation is managing dozens, if not hundreds or even thousands of models, it can become difficult to maintain the quality and integrity of the underlying data.

As Abercrombie notes, the stakes for this might be very high indeed. “Think about robotic-assisted surgery,” she said. “If there aren’t the proper constraints and management of the data, what’s to say you couldn’t cut a hole bigger than a person can handle, because the AI “sees” cancer material that is significantly larger than it actually is?”

Another challenge that we discuss on the podcast is the structure of teams within the organisation, and how, particularly with regards to larger companies, oversight into the applications being developed is too siloed. According to Abercrombie, with too many enterprises there’s a lack of consistency in processes and company-wide oversight and policy across those teams.

One of the key steps that is being overlooked in the rush towards AI, Abercrombie notes, is data literacy. Organisations and individuals need to redouble their efforts to truly understand data first. Because without that, the ethical application of AI is always going to be a difficult question.

For more deep insights into the thinking that is driving ethical AI and how enterprises are thinking about it, tune into the podcast!

Enjoy the show!

Find out more about Cortnie’s book at Amazon

Thank you to our sponsor, Talent Insights Group!

Join us in Sydney for OpsWorld: https://www.datafuturology.com/opsworld

Join our Slack Community: https://join.slack.com/t/datafuturologycircle/shared_invite/zt-z19cq4eq-ET6O49o2uySgvQWjM6a5ng

What we discussed:

00:00 Introduction

AI is a powerful tool, and as enterprise and government find more sophisticated ways to leverage the technology, there will be untold benefits returned to customers. At the same time, the responsible use of AI is of significant concern to the global population, and people are watching how its use is regulated closely.

On this week’s Data Futurology podcast, Felipe Flores presents an update on the status of regulation across Europe, China, and the US, and poses the question about whether AI regulation needs to be a global, rather than regional response.

Perhaps surprisingly, China’s taken the lead in regulating how business uses AI, Flores said. “The regulation says that businesses must notify users when an AI algorithm is playing a role in determining which information to display to them and give users the option to opt out of being targeted. The regulation also prohibits algorithms that use personal data to offer different prices to different consumers. It is really interesting that China moved early.”

Meanwhile, in the EU, the drafted regulation would categorise AI applications into one of four “risk” profiles, with oversight and accountability being scaled in kind. And in the US, much of the focus around regulation at the federal level is concerned with the potential for discrimination, while states are being left to develop their own broader frameworks.

Australia, which doesn’t yet have regulation, does have an ethical framework, which is an indication of where future regulation might go. Flores runs through that framework in this podcast as well.

For an in-depth look into the exciting and dynamic discourse around AI regulation across the world, tune into the podcast!

Enjoy the show!

Thank you to our sponsor, Talent Insights Group!

Join us in Sydney for Ops World: https://www.datafuturology.com/opsworld

Join our Slack Community: https://join.slack.com/t/datafuturologycircle/shared_invite/zt-z19cq4eq-ET6O49o2uySgvQWjM6a5ng

What we discussed:

00:00 Introduction

2:05 Discussion around AI regulation and how should different countries tackle it.

2:50 How have the US, China and the EU approached this.

5:00 EU regulations

8:05 USA regulations

9:40 Thoughts and comparison on the three approaches.

11:10 What’s happening in Australia.

Quotes:

· In March 2022, China passed a regulation that governs companies and their use of AI. The regulations applies to online recommender systems. They say the AI needs to be used in ways that are moral, ethical, accountable, transparent and that disseminate positive energy.

· Companies (in China) are expected to submit their algorithms to the government for review when they are being used at scale.

· The EU separates the ways AI can be used into four bands according to the risk involved. They have minimal risk, limited risk, high risk and unacceptable risk. The unacceptable risk covers things like social surveillance, facial recognition, etc.

· The US congress enacted a National AI Initiative Act, focused on improving research development, understanding AI and having an AI strategy within the country.

Today on the Data Futurology podcast, we have Ann Sebastian, Senior Data Scientist at Wesfarmers OneDigital, as a guest on the podcast.

As Sebastian says, she is “in the trenches” building data science products. One of her key projects in recent years has been OnePass, a subscription service that provides free delivery and other services across a range of Australia’s top brands.

“It’s an incredible experience to be part of a journey, developing an idea through to proof of concept through to production ideation to a system that is adopted across the organisation,” she said. Through the podcast, Sebastian offers some key insights into that process via some of the projects that she has worked on over the years.

Sebastian also spoke about how data science teams can be built, and how a culture of innovation can be structured within them. For just one example of this that she shares on the podcast, in her current role there is a focus on learning and development, which manifests as 10 per cent of each person’s work time being dedicated to research activities.

For her part, Sebastian is currently using that research time to work on multimodal product classification, she said. “Given the fast-moving nature of retail catalogue, and need for us as a division to form a unified view across all our divisions products, there is business significance for this research project.

“We then have fortnightly quick check ins to discuss the progress on our research projects, and that really helps us to learn from each other. This is one way that data science is embedded into our day to day in a way that makes it more real for us.”

For these insights and more on how data science products are built and evaluated, and how data scientists can be motivated and innovative within their careers, tune in to the full podcast.

Enjoy the show!

Click here to learn more about OnePass

Thank you to our sponsor, Talent Insights Group!

Join us in Sydney for Ops World: https://www.datafuturology.com/opsworld

Join our Slack Community: https://join.slack.com/t/datafuturologycircle/shared_invite/zt-z19cq4eq-ET6O49o2uySgvQWjM6a5ng

What we discussed:

0:00 Introduction

3:00 Ann talks about her experience and her remit at Wesfarmers

7:44 What are some of the use cases you’re proudest of?

15:12 Ann shares more information about her favourite use case and how it evolved from an idea to the start of the technical work.

17:58 How did you measure business impact?

20:14 How did you operationalize the models?

24:00 Can you describe your current role?

31:23 What’s your advice for people wanting to get into data science?

Quotes:

· Business teams can sometimes view data science as a mythical creature, so I love working with them to demystify data science and achieve business benefits through it.

· The use case that I'm proudest of is the automation of the complaint classification, where we implemented various natural language processing models to predict the category of the complaints using real time models.

This week on the Data Futurology podcast, we have three special guests to share insights on how data works in retail settings. Nick Merry, the Head of Analytics at flybuys (Loyalty Pacific), Kathryn Gulifa, the Head of Data and Analytics at Catch, and Stuart Garland, the Director at Talent Insights Group, join us for a wide

As Merry also notes, the days where the data team would be separate from the other lines of business are largely over. Now, the digital team is integrated into everything from marketing to security and governance, and people on that team need to be able to have conversations across all of them. “Having digital analytics, not as separate functions, but more integrated with the broader view, is one of the encouraging things that I’m seeing,” he said. For more deep insights from these three thought leaders on the changing dynamics of work in data and analytics, tune in to the podcast!

Quotes:

I’m not a fan of the data translator role because I feel it absolves data analysts from developing the skills of consultation and defining a problem. What differentiates good analysts from really good analysts, is understanding the business context and the ability to drill down into what's actually important to the business.

When it comes to recruitment, I always think the technical skills can be taught if you've got the technical aptitude. The technology landscape is changing so rapidly, all the time, that if you really try and peg yourself to recruiting people that have experienced only with particular tech, then you're really limiting your options. I think what you should be trying to find people that have not necessarily the polished and ready to go consulting skills, but the curiosity, the engagement, the wanting to understand why they do something, and what impact their work actually has on the business that they work for.

Considering people with longer or shorter tenures depends on what the role is and what you want from that individual. If you're in the process of building a platform and bringing in a data engineer that has gone across three or four different builds over the last four or five years might be useful because from that perspective, you've got three or four different pain sets, lots of experience in regards to what went wrong and, more importantly, what went right.

This week on Data Futurology we answer a burning question that is asked in the data space a lot: just what makes a good data strategy?

As we discuss, there is no one-size-fits-all approach to data strategy that will work for all organisations. This cannot be approached like a templated “best practice” to business. Instead, there are three factors to consider when devising the data strategy that will work for your business:

1) Defining where the organisation is on its journey today. How a business just starting out with data needs to approach strategy is different to an organisation with a mature data practice.

2) Deciding on where the organisation wants to get to. This refers to the need for organisations to get buy-in across the business to a data strategy. Without that alignment the project is prone to failure.

3) Developing the execution path, to take the organisation from where it is now, to where it is going to be. The better defined this pathway is the more likely it is that the project will stay on-track and on-goal.

What distinguishes a good data strategy is one that is aspirational in nature. “Aspirational” doesn’t mean that the goal needs to be futuristic or difficult to achieve. It could be grounded and realistic, and simply an effort to step up from where the organisation is currently. But having that clear goal and a vision for the value it will deliver to the organisation at the end of the journey is critical to motivate the effort behind the data strategy.

In this in-depth discussion, we lay out the approaches and use cases that motivate successful data strategies and highlight how organisations can approach each stage of the journey.

Enjoy the show!

Thank you to our sponsor, Talent Insights Group!

Join us in Sydney for Ops World: https://www.datafuturology.com/opsworld

Join our Slack Community: https://join.slack.com/t/datafuturolo...

0:00 Introduction

1:09 How do you devise a data strategy? What sets apart the good from the bad  in a data-driven strategy?

3:30 Data strategies encompass everything, they are broader than analytics, AI and tech strategies.

5:10 Get alignment on where the organisation is today and where it wants to be  in the future.

7:45 The importance of having a prioritised set of use cases for your data strategy.

10:05 The four components needed to help you prioritise data use cases.

12:50 The two sides of organisational readiness.

This week on the Data Futurology podcast, we have the special privilege to host Dr. David A. Bader, a Distinguished Professor at the New Jersey Institute of Technology, and the inaugural director of the Institute for Data Science there.

Bader joins us on the podcast to discuss massive graph analytics, a topic that he is a recognised expert in and has recently published a book on. He and his team are currently working on a project that will allow anyone, via the Jupyter Notebook and Python, to leverage their data science framework, running on “tens of terabytes” of data. “It is quite exciting to democratise data science – and especially graph analytics – so that anyone with a problem that knows Python can work with some of the largest data sets,” he said.

According to Bader, graphs are now a mainstream part of data science and a way to solve the most challenging and complex problems in the enterprise. “A graph abstracts relationships between objects, and any problem that we can abstract where we have relationships between objects, we could use graph analytics to solve,” he said.

Much of Bader’s work – including through his book – is focused on helping organisations grapple with the exponential growth in data, and the impact that this has on their ability to dedicate adequate resources to work at scale. As he said, being able to do that is going to be fundamental to humanity’s ability to respond to the many real challenges that it faces ahead.

“I want equitable access for everyone to be able to work on these problems, and to find new discoveries that are important, and help solve global grand challenges,” he said. “I think that we have many issues in the world today. And if we give more capabilities to those with data, and let them empower the data will make the world a much better place.”

For more deep insights on the importance and value of massive graph analytics, tune in to our conversation with Dr. David A. Bader.

Enjoy the show!

Thank you to our sponsor, Talent Insights Group!

Join us in Sydney for Ops World: https://www.datafuturology.com/opsworld

Join our Slack Community: https://join.slack.com/t/datafuturologycircle/shared_invite/zt-z19cq4eq-ET6O49o2uySgvQWjM6a5ng

Read the full podcast episode summary here.