Hypercube Podcast Transcript

Title: Data Science Behind EV Charging Infrastructure

Host: Adam Sroka
Guest: Ali Saad

Intro: Welcome to the Hypercube podcast, where we explore how companies in the energy and utility sector leverage data analytics to make smarter decisions and accelerate business growth. I’m Adam Sroka, founder of Hypercube, a strategic consultancy that supports asset owner-operators, traders, route-to-market providers, and energy services companies to unlock the power of data.

If you’re interested in hearing real-world examples of how data and AI are advancing the energy sector, this is the show for you. 

Adam Sroka: Hello, and welcome back to the Hypercube podcast. Today, I am honoured and humbled to be joined by Ali Saad, who is Data Science Lead at Connected Kerb. Ali, could you introduce yourself and give the audience a little bit of clues to what you get up to over at Connected Kerb? 

Ali Saad: Absolutely. Thank you, Adam. And thank you for having me.

Yeah, I’m the data scientist lead at Connected Kerb. I’ve been with the business for over two years now. And Connected Kerb is an AV charging company in the UK. We are based in London. I think we are second or third largest network in the UK. My job with Connected Kerb, you can think about it from a data science point of view, starting from data analysis, data modelling, and so on.

But I’m also responsible about the team. We’re internally for insights and data science and for the data infrastructure when it comes to modelling, machine learning, forecasting, and so on. 

Adam Sroka: So yeah, exactly right. And Connected Kerb, you’re in that really crunchy problem space of like, how do you get charging points distributed across the network, across geographic locations?

You’ve got different legislation, ever changing, you’ve got grants and things like that. I’d be interested to know, yeah your thoughts on the upcoming challenges and how like data and AI is really helping in that journey.

Ali Saad:  I love this question. All right. So I’m thinking about different aspects in here.

So you mentioned deployment. And you mentioned data. So let’s start with this internally. We have that tool called site selection tool and what we’ve done into this tool. Basically, connecting data, do you data from different sources? You can think about starting from a simple as such as data to maybe parking lots and.

On street parking, all of this info, and we added this in a way where this info should help us make a decision on what would be a good location to deploy a charge point. And something for the audience, we at Connected Kerb mostly deploy charge points on street for the public. So it is not very limited to parking lots or clients.

It is for the public. So somebody without a driveway, we can go install something on their roads and they can just park there and George and with this tool, we managed to be able to understand what is happening with the street. So we started learning from the current charge points and buttons on different geographical region and we will be able to select what we call the best or optimum location to the bear charge point with considerations to things like maybe flooding areas or maybe Other considerations that we might want to take internally.

So away from site selection tool and deploying charge points, the most interesting aspect for us would be grid balance and energy, how we’re going to do that, and it is not hot topic for us only. I think it is for the grid as well, for multiple providers, for multiple stakeholders, because when you think about this, let’s assume now, We have 30 million cars or EVs on the route and they need to charge.

So imagine all of them going to charge, and this is some, just a simple speed, seven kilowatt or four kilowatt. It’s very simple, but this is a huge amount of energy. A huge demand might come at any moment in time. And when we think about the future, you want to balance the grid. And from that point of view, You want systems that are capable of predicting these patterns, predicting these behavior, and being able to interact with them in an automated way without users interaction.

And from that point of view, there are multiple opportunities for us as charge point operators to balance this and to contribute to the grid. And there’s, there are several technologies. You have vacant to grit, which is a trial, a card can produce energy back to the grit and can support the grit. And you have other things like smart charging, where you can plug in and discover would tell.

Would stay plugged in, but it wouldn’t charge at the moment. It would charge during night or something where there’s no load in the grid. And these are the technologies that we need to develop and enhance internally in order to provide that back to the user. Other aspects of these would be education. So we want to educate users.

And how these things works from an internal point of view, I have to be able to predict that user behavior and to predict it in a very good way where the user barely notice a change. 

Adam Sroka: And that’s a big change. So that makes me think like you must be dealing with lots of time series data, lots of open and private data sources and things like that.

Are you finding access to the right? Data, but also skills to, to process and wrangle quite a large, complicated data set. You find that quite easy to get hold of in your team. 

Ali Saad: To be honest, no, at all. We’re always craving for high quality data. And actually funny thing I was working a couple of weeks ago on data modeling for cost energy for costing.

And it’s a type series data, as you said, but at the same time, one of the questions that I was asking for, yeah, we know how much energy is being delivered by each location, but at the same time, you want to Why? And why could be delivered by there is a shopping mall nearby or there is a business park or there is whatever other geographical things you have within this location.

And if you have this data, if you have this rich, as rich as could be, it would enhance your modeling. If you have traffic data or events data, so imagine that there’s a concert nearby that’s happening today, So you should expect some demand on these locations nearby. And if you have this data in an accurate manner, you would be able to predict this in a much more greater way.

And that’s very complex and hard to get. An d even if you manage to access data or find high quality data, the cost of acquiring the data And being able to make a profit from it, it would be very much harder. Other data points, which is in, I would say for not just for us, but for every charge point operator would be the internal systems that when you think about a charge point, it’s an IOT device.

So it is connected to our back office through wifi. And at any moment, that connection could be interrupted. or anything could happen with a chart point from a mechanical point of view or hardware point of view. And at that moment, it wouldn’t work. And you want to predict these kinds of events.

Sometimes it is predictable. Sometimes it is not, but at the same time, we know when you get out loads of these events, you still be able to predict something or if it is raining and flooding areas. So all of these, there’s no system at the moment. That tackles all of these challenges. You will find some system might just use previous energy consumption to predict next energy consumption.

And that could be an indication level of something. It’d be as simple as this, but when it comes to complexity and having much more greater outcomes or. So predictability, especially for these abnormal events, it is not there yet. And that is one of the great challenge that we always paid for. 

Adam Sroka: It’s really interesting because I always think like optimization is a bit of a dirty word in the energy sector because for some people that have done like Optimization at university and things like that.

It’s one thing and it’s this complicated machinery of doing things a certain way, but then a lot of the time in the industry, optimization means sometimes we’ll turn it off. Not all the time, but sometimes we’ll just turn it on and off. And it’s very binary. And you think there’s a long way to go.

You talk through a lot of different use cases there. You’ve talked through a few products and things, and I guess you’ve a lot of experience with startups and it bumps into that startup problem. How do you pick what? Data challenges and like data products and things to develop, like what, how do you drive a strategy for your data science team to tick off the most valuable use cases because you could get pretty lost, right? In what was possible? 

Ali Saad: This is a big problem and I have two things usually a sit on the back of my head, a very exciting project that I might be able to see a value in it, but everybody doesn’t have the same vision. And the other aspect would be a project that everybody in the company knows if we manage to do it right, it will enhance our operations or reduce costs or increase profitability.

And usually these projects are linked directly to the financial model. So it’s about what your business is facing at the moment. And you go from the business challenge down to the data science. Hey, This is the most significant problem because if you manage to enhance utilization by 1 percent or 2%, it wouldn’t be, we would be able to make this amount of money.

And this is how we pick problems. And usually these are the ones that we work on for a long term, but also there are some problems where you might think about the time, the schedule around it, and you might be able to fit in your time and calendar and happily in Connected Kerb, we had several initiatives with these little projects, little mini projects, the company found it very interesting that became a significant part of other products.

So it’s how you, as a data scientist, see the world net. Because nobody see the wallet like you and how the company itself, especially from a financial point of view, see the wallet and going backwards to the new strides.

Adam Sroka: So in your role as data science lead, then I guess like it might be useful to just hear what does a typical day feel like?

Are you hands on coding most of the time, or is it actually mostly that steering the ship and helping identify use cases and write the business case and so on and so forth? 

Ali Saad: That’s a good question. And in the current moment, the person who is building, but also I have to shape things, so I’m trying to balance between them.

That’s because we’re quite a small team at the moment. In the future, I want to direct the ship in different ways because there’s always something you think about and you find very interesting, but yeah, we balancing between both. 

Adam Sroka: Okay. And I guess, what do you see as the biggest blockers or like barriers to, to scale like those, are there any kind of golden nuggets that you think if you could just do that one thing, it would accelerate really quickly that you can share with us?

Or is that all secret sauce? 

Ali Saad: No, I think I can share it. Yes. When it comes to hiring and your team. I always think about it in a way where I’m not shooting for the optimal team. I’m shooting for the most efficient team or optimal team. So not a dream team. I have four, five, 10 people working with me now.

I want the number of people that I need to achieve this project or this goal by end of year or whatever period of time. And I’m thinking of it in that way because of multiple aspects. One big aspect would be the current industry and the current market. It is quite in risk. So you want to balance what you do and your spendings.

At the same time, I want to think about the future of the projects that’s and the impact of work that I’m going to do. If I’m going to hire three, four people and I need a budget of half a million or something a year. Okay, great. But you need to be able to make us 5 million or 10 million. So whatever the number is, and if I can’t see this on the balance sheet, I wouldn’t be happy.

And this is one of the unique things I always think about myself. I always think of myself as a kind of a, between business and technical, where if I’m running that business, I would I do that money or would I find somebody who can act like a joker with a bit of a skill on the data engineering or data science and can do multiple things.

And that strain enables me to think about the skills that I need. of whatever job or task I need to hire for. So I want to ask you a kind of double part question off the back of that. Both at Connected Kerb, but generally as advice for startups. So they could be different answers. Do you prefer like a team of specialists that work together well, or a team of generalists that can be a bit more flexible and do a lot, but maybe not be as effective in their specialist domains?

Ali Saad: I love that. I would say it depends. It depends on the stage. 

Adam Sroka: You can tell an answer. I like it.

Ali Saad: Yep. Statisticians usually, their typical answer on any statistical question is I don’t It depends. And I think to be honest, away from that, I’m serious about this. It depends if I’m starting. So at Connected Kerb, I remember myself reading that book, How to Build Data Driven Culture.

So I’m remembering that name correctly. It is by Carl Anderson. It is one of the greatest books that helps you think about how to design a communication line and how to design a team within businesses. And I like his framework of having. Mixed methodology between having a central environment or a central team where everything can start from there and spread across different teams or having a specialist team.

So a team of data engineers, a team of data scientists, a team of data analysts and so on. And when I think about that, I always think about having a central team, especially when you start. They can work together. That central team consists of three or four people. That’s a data analyst, data scientist, data engineer, and maybe machine learning engineer.

And that team would enable you to build your architecture and the frame that you want to work with and the communication line. So your data analysts are your frontier line who communicates with different functions within the business. And whatever thing you do, even if it is machine learning modeling, That needs to be presented through them because they can visualize things in a nice way and they can communicate things in a nice way.

And I found that very effective when I designed it in that way in Connected Kerb and later on that data analyst team can communicate with what we call data professionals in different teams. So we have somebody in, for example, in finance or in marketing or, Operations who can read and understand data, and these would be not in our team, but they do understand their context of business, what they do on a daily basis and understands data and they can communicate that with their teams.

So that’s a data analyst and another other data professionals. And then we have a data engineer or data scientist. So the data engineer can build the foundation of data pipelines and other stuff. And then the data scientist can work or build on that, on the top of that, build machine learning models, casting models, and so on.

Adam Sroka: So I’m going to ask a follow up question off the back of that. How important do you think industry experience is in, in those hiring decisions and when you’re picking a team? 

Ali Saad: You always prefer somebody with a little bit experience, but it is hard to find as well. So you always want to be mindful about that.

And I think the most critical aspect is not just the experience, maybe the topic, maybe time series analytics, if I were to do that, maybe neural nets and machine learning or deep learning, maybe something else related to the technology that we’re using. If you’re using Databricks or Snowflake or any other types of technology, they have that limited expertise.

But the most important aspects for me would be the foundation to be right. If you’ve got the foundation right, and you’ve got that relative project or similar project that you worked on, I think this is the most critical aspects for us. And when I started, I wasn’t, I didn’t have any experience on energy sector, but I managed to learn that and acquire that knowledge.

Maybe from a technical point of view, because we work a lot on time series data, that would be, at real time analytics, those would be two critical skills to look at. 

Adam Sroka: Just changing gears a bit then, would you have a favourite tech stack or have you got favourite technologies that really drive the delivery at Connected Kerb?

 

Ali Saad: Coming from different cloud providers, different solutions, you find it fairly easy to learn a new technology. It takes a bit of time, yes. But it wouldn’t limit you from hiring somebody. If you find somebody that is really good at something and by the end of it, If you’re really good at building prediction models, it doesn’t matter if you sit on snowflake or data bricks or any other aspect, you know how to write the code, you know how to evaluate it, you know how to do the stuff.

So that shouldn’t be a problem, especially if you if we have somebody on the team who can pick up the work from that person. So if you got somebody who’s really good at statistical modeling, but that person use ours, for example, which is not Ingley used fairly for production, but we got a data engineer who can transfer that code into production and make it in Python and build the entire pipeline.

So why would I stop from hiring that person? And that would depends. You have to make these decisions by evaluating and thinking how the workflow And heads over between one person and the other person. There’s not much gaps in between. Now, I wouldn’t stress on relative experience. I know it would be amazing to have.

I wouldn’t stop hiring somebody if he’s good and does have a lot of experience. 

Adam Sroka: What about your thoughts on emerging technologies, gen AI and things like that? Do you think they’re going to have a significant impact on the work you’re doing? 

Ali Saad: Absolutely. It is actually one of the blessings that we currently have, but I always think about it in a way where it depends on your data architecture and how your data articulated.

If you have a really good data architecture, you can easily build your model. Or connect your GBT model into new data architecture. And you can provide the context of this data as well. And that enables you to, from any, but you’re able to say on the business who can happily interact with GPT can go there and create the data, do some modeling and do some work.

Of course, there’s always things to consider about accuracies and whatever response, because sometimes it might be accurate. I see in many cases. It can lift us, lift some work from us. So anybody can go do some analysis and build on it. And we can just go there, evaluate it and say, it doesn’t make sense or not.

And internally for us as well, when I do some modeling, I can go and ask to Agility to code some stuff for me. It’s always amazing to do that. Sometimes you have to think about it, but. Still, for me as a data scientist, I need to do some mathematical coding and evaluation. It is not that accurate all the time.

I have to revisit and rethink about it again. 

Adam Sroka: And what about the energy sector as a whole? Do you spot any trends in how the technology might be changing things there? Or is there things that you’re quite excited to see play out? 

Ali Saad: Actually, this is one of the most exciting topics for me. I have to just wait for the future.

Hopefully I would contribute more. But if you remember our conversation regarding balancing the grid and vehicle to grid, grid to vehicle, and so on, when you think about this, all what you want to consider is a system that capable of reading all of these data inputs from the grid from actual reality.

Maybe it’s from weather forecasting, maybe from events, and if you have that brain and you have this data accurate with you and you build your machine learning models on the top of that, you can predict what’s going to happen, and you might be able to take your actions before it happens. And I see this as a system does not serve one either charge point operator.

It should serve them nationwide, I would say. And that would be a super powerful system, because it is not, if you think about the current circumstances, we don’t have dynamic pricing yet. The energy itself is priced based, I think, on a region level. So if you would go one level deeper to a zone or node level, and having that with the dynamic pricing, you should be able to play in different, with a price of different way, where you can benefit the end user.

This is a very cheap price. To charge go and charge at the same time we can we would be able to make more money the grid would be able to save a lot of money because of that balancing and that what makes me excited because you can see the entire cycle and you can balance it somehow it reminds me of With flying systems, because it is flying systems, quite connected, all connected together, especially when it comes to airspace and whatever changes are going to make, it’s going to impact the price and demand and jolting and all of that.

And I see similar use case in energy, but it is, we’re not far from there. I think maybe in three, five years time, we should be able to see something tangible. 

Adam Sroka: And it’s a place that from what we see. With the work we’ve done with risk your batteries and things, it’s, there may be a little bit further ahead because it’s easier to do because there’s fewer of them, fewer players in the market.

But we’re seeing lots and lots of interest now in, in getting into that game. I don’t think there’s any one technology provider that’s won it yet. So it’ll be interesting to see when you’ve got this huge tidal wave. EVs joining the battlefield as it were, what difference that’ll make and the advances we’ll see is definitely an interesting time to get into energy and data, would you have any advice for people that are listening to this and thinking, wow, I really want Ali’s, Job, like, how do I get into  Connected Kerb or data science? Would you have any advice for young, early career professionals? 

Ali Saad: Young, early career professionals. Okay. I would highlight three aspects, technology and tools.

Plenty of tools online. Being familiar with two or three of them would be great. Databricks, Nuthlake, or any other tools. That’s one aspect. If you managed to get yourself familiar with it, that’s a plus point. The second aspect would be deployment and production, how to move things to production, machine learning model, or other aspects.

The third aspect would be relative domain experience, or not really domain, but maybe the use case itself. When it comes to time series analysis, when it comes to real time analytics, all of that. If you got these three aspects, it would be great. But also, there are other aspects that I always find it less, less trendy, I would say.

Many times when I go consult companies or something like that, I always go back to the data analyst people. And I’m asking them, how much do you know about your company and what is happening before forecasting anything? And most of the time it is not that much. And this is an area where it is underrated.

It is very critical because if you know how things are actually happening now, and you’ve got that, sphere in the front of you, you can look at it from different perspectives. You would be able to tell, okay, this is an area where it might, if we enhanced might change something, or at least some knowledge we know.

And whatever modeling you’re going to do, we would have some expectations on it. And that’s BI tools, Power BI or local or Tableau. is another aspect. It is not just about the tool, it’s about the story and how you’re going to tell a story to your audience. And I just love and these people who would go through this area and focus into it and doesn’t think about any other aspect when it comes to data science or any other aspect.

They are so powerful and I consider them as the frontier of whatever it means to me, because we communicate that to a lot of people. I 

Adam Sroka: I love that. I think you see this a lot in the Lots of techies get excited and really revved up about the technology because it is fun and it’s interesting and we all love the complexity of it and we’re solving hard problems.

But linking it back to like, why is the business doing this at all? Why are you employed? And what is the business trying to achieve with this? Isn’t always crystal clear and it, and you’re right, those people that live in those kind of boundaries between the business domain and the end users, those storytellers have to understand that and have to translate both ways.

A really powerful skill if you can learn it. So yeah, really agree on that. Constance this time. Thank you so much for today. It’s been great. Before we let anyone leave, I always like to ask if our audience want to hear more about you, the work you’re doing, or what Connected Kerb are up to. Is there somewhere we could send them or is there anything you’d like to promote or plug whilst you’re on the show?

Ali Saad: I would appreciate anybody what’s connected to me that would be efficient. You can find me on LinkedIn. This is the best way to find me. Mostly I’m active there. Usually it’s just A L Y with Y. And then number two, S A A D. If you Google this, you will find me everywhere on the internet. This is my handlers for everything.

I would be happy for anybody to connect. 

Adam Sroka: It’s been a pleasure. Thank you very much. It sounds like you’re doing some really interesting stuff over at Connected Kerb. And I look forward to hearing more about the rest of the future. 

Ali Saad: Thank you very much, Adam. It was a pleasure meeting you. And I did enjoy it very much.

Outro: And that’s it for this episode of the Hypercube podcast. Thanks for tuning in today.

If you have any questions about the topics we covered, you can reach out to us on LinkedIn or check out our website at  wearehypercube.com. You can also join Beyond Energy, our Slack community of data leaders from the sector. There’s a link to sign up in the episode description. We’re just getting this show off the ground, so if you like today’s episode, please leave us a rating, review, or subscribe wherever you get your podcasts.

It all really helps. See you next time.