GRIFFITH UNIVERSITY - RELATIONAL INSIGHTS DATA LAB

EPISODE TRANSCRIPT

Rhetta Chappell (host): Hi and welcome to Show Me the Data, a podcast where we discuss the many ways in which our lives and the decisions we make are impacted and depend on data. I’m Rhetta your host for today, and I’m a Data Scientist & Partnerships Lead at Griffith University.

On today’s episode of Show Me the Data, we’re thrilled to welcome Saeed Barghi, a seasoned expert with over 15 years of experience in crafting and constructing data platforms across diverse industries and harnessing technologies to improve decision making. Since starting at us in 2021, Saeed has been dedicated to the public sector throughout the Asia-Pacific and Japan regions. In his daily endeavors, he plays a pivotal role in empowering AWS customers to embrace a data driven approach to work. Let’s jump right into our conversation. Show Me the Data acknowledges the Jagera peoples who are the traditional custodians of the land on which we are recording today. And we pay respect to the elder’s past, present and emerging.

We’re so happy to have you here, Saeed. So thank you for taking the time to speak with us. We’d love to hear in your own words, to hear you kind of explain your role, how you got into data, what your key priorities are, and I guess your intended impact at AWS.

Saeed Barghi: So I started actually my career as a web developer, but soon I realised I can make things work. But my brain is not wired to care if they look pretty or not. And that’s not a good quality to have as a web developer. So gradually I started working more with databases, with ETL processes, and I realised that’s something that I’m good at and that’s something that I’m passionate about. So I’ve been working with data for the past 17, 18 years across different organisations, different industries, and with different technologies. I joined AWS just over two years ago and I cover public sector across the APJ region and what gets me going every day is to innovative ways that my customers use their data on and how they use it for the public good.

And because I cover a very wide geographical area, what I learned from one country and I can teach all my other customers about it. It’s something that I really cherish in this role, and that is how I make the biggest impact.

Rhetta: Because you’re working with all these different types of organisations and you’re kind of helping them to make the most of their data on a daily basis. You’ve probably kind of seen all levels of data maturity and seen really cool, amazing things. What do you think it means to be a truly data driven organisation?

Saeed:
So if you asked me this question a year ago, I would have said being data driven means making sure the right data is available to the right person at the right time so that they can make the right decision in response to internal or external events. But we all know that with large language models and generative AI, machine learning and AI is much more available to a lot of different person organisations now. So I would like to add another dimension to that definition, and that is to make sure the right machine learning model trained with the right training data set is available to those decision makers in the form of some sort of application that help them and act as their digital assistant to help them again make the right decision. And this doesn’t mean that every business decision maker needs to become a data scientist or develop those models. What I’m referring to is a set of applications that interact with those pre-trained models via a set of APIs or other interfaces, and again, use that as a digital assistant. And that’s why a lot of our customers have already started using one of our newest services to build that machine learning into their application that they develop and they release it to their customers or internal decision makers.

Dr Tom Verhelst: Your extension of that definition of a data migration is quite interesting because it has multiple components. Its infrastructure is data, it has AI or logic, and then you have applications. Saying that, where does AWS see itself, is it in one of these segments or does it see itself as a provider of infrastructure, a provider of data and insights, provider of AI and algorithms and then sort of the host of these applications or all of the above?

Saeed: The short answer is all of the above. We at AWS offer the world’s most comprehensive and broadly adopted cloud infrastructure. We have over 200 fully developed and fully featured services that our customers can use to deliver different type of use cases and those sort of services enable our customers to build and develop fit for purpose and future proof data platforms. And one of the most important characteristics of this data platform is that they are modular so they can add different layer, different components to them or take away some and expand it for their new use cases or expand it to cover more users. And on the AI and ML services, we help our customers to gain deeper insights from their data efficiently and cost effectively by adopting machine learning at different levels of the organisation. I talked about how we can put AI and ML into the core of application development, but another example that I can provide here is that, for example, if an organisation wants to enable their data analysts to use more AI and ML in AWS they don’t need to spend months or years to learn about data science and developing those models, they can use SQL, which is the most widely used tool for data scientist to interact with those models and enable again those data analysts to incorporate the machine learning models into the analytics that they develop. And what one direct impact of this would be that because they use SQL to interact with those machine learning models, they can very easily incorporate them into reports and dashboards. And instead of just showing what’s happening within the organisation or within their industry that operating, they can extend a dashboard and show of what the current trends is going to lead to in short term and, you know, long term future.

Rhetta: Given that A.I. models require such a vast amount of data, which is something that you’ve talked on for their training and for the inferences that they make, how are companies like AWS and organisations you partner with in the different government sectors working to keep citizen data anonymised, secure, and also protecting it while still talking about leveraging all these amazing benefits that you’ve just spoken to?

Saeed: So security is job zero. I guess what it means is that we provide our customers with all the tools and frameworks necessary to secure the workloads on the cloud. And security on the cloud has two dimensions to it. The first one is security of the cloud that is where Amazon Web Services is responsible for making sure that the infrastructure that runs our customer workloads is secure. For example, no unauthorised user gets access to our data centers. And the second one is the security in the cloud. This is where our customers need to make sure that their workloads on the cloud are secure according to our standards. By configuring the services that we provide to them according, again, to their standards. And here we help them to achieve that level of security as well. For example, if they bring their data into Amazon S3, which is a service that is used to build data lakes by default, all the public access to the data stored in Amazon S3 is blocked. So when they bring in new data or when they create a new bucket by default, nobody from internet will be able to get access to our data. If there are some cases that they need to open that data to the public. For example, if they are hosting a website on Amazon S3, they need to go and explicitly change that.

The other example that I want to mention here is that when it comes to data masking if enable our customers to statically mask their data. What I mean by that is to mask the data on the fly when they moved from one environment to another so that it is already masked and encrypted before it hits the storage layer on the target. And we also provide them with dynamic data masking, and that is when, according to the role a user plays in the organisation, when they query the data set, they see different views of that data stored in Amazon Redshift, for example, if there is credit card information stored in a table only the payroll agent or payer, those who are working payroll will be able to see the full credit card number. Everyone else will either see the completely masked version of it or they will be able to only see the first four digits.

And I want to add here that our services that allow our customers to store data, they allow them to encrypt that data with their own keys or AWS provided key as well. So it’s fully in their control on how they want to implement the security and privacy controls and how to approach it. That best match of the standards that they need to follow.

Tom: So given the unique challenges and opportunities of the Australian public sector, how do you see AWS solutions, especially AI driven ones, transform public service and operations? And can you share any real world examples or use cases where the integration of AI and data has led to significant improvements in public service delivery or operational efficiency of a department?

Saeed: So the very first example that I want to talk about here is something that I’m very passionate about is about global warming and climate change. I refer to detecting wildfire smoke. So one of the effects of climate change, such as extreme heat and long drought, has been the increased risk of the extent of wildfires. Wildfires, in turn contribute to climate change through greenhouse emissions. Right. And in terms of the impact, in addition to the risk to the wildlife and biodiversity, wildfire can destroy and disrupt property and infrastructure as well. So early detection is very critical in these cases. We hear that in some cases the size of a fire, depending on the wind and the area that is happening can double in size every 3 to 5 minutes. So a solution that is actually used in the United States right now, in Amazon Web Services is that they automate the solution that looks into the video feed of captured of certain areas. And if there is a fire breaking somewhere, the solution, automatically analyses that feed, extracts the information such as the location, the size and the direction at which is growing, and sends that to the first responders. And I guess this is an area that public sector and private sector can work on together because it has benefits both for public sector through protecting the wild life and biodiversity and for private sector to protect their assets. And you know, what they use to offer services such as, for example, for utility companies to their customers.

I have a couple of other customer examples here as well. Something that I have been working on recently is around improving the citizen experience. And that can happen in different shapes and forms. The first one is through the agent, the chat bot that are equipped with LLMs. Those chat chatbots usually provide better answer to the customers, so they reduce the need for citizens to have to call the call center and talk to a human. And because they are 24 seven, if I, for example, I have some free time before going to bed, I can ask my question and do what I need and don’t have to wait for the next day. The other use case here is on improving the efficiency of the call center in case someone needs to talk to them. And this comes in the picture when an AI and ML capable application will go through the transcripts of an agent’s conversation with the customers, summarises that and store that in application so that the agent doesn’t need to spend time on typing down the summary and, you know, pass it on to the next agent who is going to probably take over this case and they can talk to more customers and they can answer more questions.

Tom: In the context of the Australian Public Service, where budget is always constrained. How does AWS ensure that its AI and data solutions remain cost effective while still delivering on the performance requirements? Because what like what you described before with the satellite or with the imagery, if it’s potentially satellite imagery, that’s quite a high volume so that the data that comes in is high volume. So we have quite a bit of storage and then the compute is real time imagery. So the compute requirements will be like it would be quite expensive to be running that for large areas 24 seven.

Saeed: So the different aspects are coming to picture for cost optimisation and cloud first of all, moving to cloud on its own helps companies save cost. How? Because they benefit from a massive economy for scale. And what I mean by that is that the fixed costs of running cloud infrastructure will be distributed across more customers and every customer will pay a smaller portion of it. The only thing I want to mention here is that on AWS, whenever there is a cost saving on running our services, we propagate that to our customers. That’s why since the start of Amazon Web Services in 2006, we have reduced the price of our services over 120 times. The other very important factor here is that on the cloud, customers and organisations only pay for what they use for they don’t need to go and provision a massive amount of resources just in case there is a fluctuation. And they need, additional processing power at some point in the future. And as an example of some of our services, I want to mention AWS training and eight in first year, which are designed to deliver higher cost saving for deep learning applications, and they offer the lowest cost for training models and running inferences on cloud.

Example I can very quickly mention here is our data warehousing solution and that offers up to five times better price performance than any other cloud based data warehouse. And when it comes to building data lakes on the cloud, which is very popular these days because of, as you mentioned, there’s a huge amount of data being stored on cloud. Amazon S3 enables our customers to automate the process of moving the data that is not needed actively for querying and analytics to their cold, their storage layer. And therefore our customers will pay less for that.

And in addition to that, we have AWS Well-Architected framework, which is what we use to go through the use cases and of workloads our customers run looking to their ambitions and future use cases and offer them where they can save cost and how they can design their solutions so that the solutions are always cost optimised.

And just to close this conversation, I want to mention that the key word here is an optimisation. Cost optimising doesn’t necessarily means lowering the cost, but it means being able to prioritise the use cases that are very important for the organisation and put enough resources on them to make sure that they are successful. And those kind of decision making should happen in collaboration between technical and business stakeholders. No organisation wants to adopt a new technology or a new framework just for the sake of playing with it. There should be a very good match and, you know, lineage between what the business decision makers and the customers of the organisation need to technology days adopt it, and it’s used by that organisation to deliver them.

Rhetta: And I’d like to just quickly circle back to something you mentioned and it was around improving that citizen experience through these kind of AI powered chat bots. What I would be curious about is how you kind of see the future of these evolving. So currently with platforms that have these built into them, there’s like a heavy reliance on the users putting in prompts. And so they have to be able to pose engaging, pertinent questions. Do you see the future of that evolving like the other side of the coin? Like how do we, make them more accessible and even increase their overall utility in the future? 

Saeed: So those chatbots can be customised, right? So if for example, if our organisation is mainly offering financial services, they chat what can be already customised to look at the question with that lens. So that is one way chatbots can be customised. For example, asking the customer with start the chat to answer a few very short questions to set the context for that chat. And that will be passed on to the machine learning model to find the right answer according to that context. 

Rhetta: Yeah, that’s super interesting. And I guess that leads us to our last question, and I think this is one that we ask all of our guests, but it’ll be quite interesting to hear what you say because as you’ve just explained, you’re really kind of about helping other organisations make the most of their data. It’s not that you’re bringing data to help them understand their particular problem or whatever it is that they’re working on. But is there a data set that as you kind of work across this APJ region, working in the public sector, trying to drive these positive changes, is there one data set that you could have access to that would kind of unlock some fruitful insights, then help the work that you do and potentially explain why that would be?

Saeed: So wherever I go, whichever country they go, I hear about how difficult it is to find the right talent, especially when it comes to AI and ML and data analytics. And I’m originally from Iran. I still follow the news from that part of the world. And one event that really broke my heart was the takeover of Afghanistan by Taliban and the fact that older women had to go back, to go back home to university professors, doctors, engineers, and they’re not allowed to work now.

So I would like to be able to understand what is the cost of bringing some of them to countries like Australia, where we are very good at, you know, accepting migrants like myself, educate them and, you know, help them understand how it what it means to work here and the business impact and the economical impact. They’re going to have in Australia in the longer run. Is that something that we can look into? Is that something that, you know, if there’s a cost benefit sort of equation to it and how that will play out, that is something that I would like to hands on.

Rhetta: That’s a really beautiful and heartfelt answer and very unique. So thank you and thank you so much for sharing your thoughts and your time with us today, Saeed. Is there anything else that you’d like to add to kind of let our listeners know that you want them to understand about you or the work that you do?

Saeed: I just wanted to mention that. And if customers are using AWS we have different types of professionals, different types of specialists that specialise in different areas, if they have questions about data analytics, security operations, we have people that can help them out. So always reach out to your accountant, always reach out to your contacts and and they really put you in touch with who who can help you the best.

Rhetta:  Thank you so much Saeed and we’ll make sure we put your details in the show notes as well so people can reach out to you if they want to find you.

Saeed: Thank you.

Rhetta: To listen to more episodes of Show Me the Data, head to your favorite podcast provider or visit our website ridl.com.au and look for the podcast tab. We hope that by sharing these conversations about data informed decision making, we can help to inform a more inclusive, ethical and forward thinking future. Making data matter is what we’re all about, and we’d love to hear why data matters to you. To get in touch, you can tweet us @G_RIDL, send us an email or better yet, follow, subscribe and leave us a five star review. Thank you for listening. And that’s it. Till next time.