00:00:05 The role and responsibilities of a supply chain scientist.
00:00:31 Maximilian’s background and how he joined the company.
00:02:03 The company’s reason for implementing the role of supply chain scientist.
00:04:14 The three components of the role of a supply chain scientist.
00:07:35 The priority of a supply chain scientist in the morning.
00:08:02 Balancing communication with clients and implementation work.
00:08:44 Difficulty with the classical approach of having separate roles for communication and implementation.
00:11:02 The advantage of being the single point of contact for clients and having direct implementation role.
00:12:54 The challenge of being accountable for multiple roles and stakeholders.
00:14:25 The difficulty of converting data scientists into supply chain scientists.
00:16:00 Discussion about the progress of PhD students on the software engineering and data science tracks.
00:17:07 What is rewarding about the role of a supply chain scientist.
00:18:16 Explanation of the diversity of the team at Loca and why it wasn’t intentional.
00:22:48 Max’s advice for someone considering a career in the supply chain industry.
00:23:00 Joannes’ advice for aspiring supply chain scientists.

Summary

In this interview, Kieran Chandler discusses the role of supply chain scientists at Lokad with founder Joannes Vermorel and Supply Chain Scientist Maximilian Barth. Vermorel explains that traditional data scientists were insufficient, prompting the creation of the supply chain scientist role. Barth shares that supply chain scientists focus on technical, relational, and project management aspects. Lokad’s unique approach involves supply chain scientists interacting directly with clients, eliminating middle managers. Vermorel and Barth emphasize the importance of practical problem-solving, hands-on experience, and open-mindedness for success in the supply chain industry. They also stress the value of a diverse workforce, prioritizing skills and competencies over nationality or gender.

Extended Summary

In this interview, Kieran Chandler, the host, discusses the role and importance of supply chain scientists at Lokad, a software company that specializes in supply chain optimization. He is joined by Joannes Vermorel, the founder of Lokad, and Maximilian Barth, a Supply Chain Scientist at Lokad.

Maximilian Barth starts by sharing his background and how he came to join Lokad. He mentions that he is German and has lived in various countries including the US, Finland, and Australia before moving to France. Like most supply chain scientists at Lokad, he has a STEM background, but his expertise is in finance. Barth highlights the similarities between finance and supply chain management, as both involve optimizing for maximum return while minimizing exposure to risk.

Joannes Vermorel explains the reason behind implementing the supply chain scientist role at Lokad. Initially, the company tried to work with traditional data scientists, but it proved to be ineffective. Vermorel includes himself in the first round of inadequate data scientists, as he was working on computational biology and distributed machine learning at the time. However, he soon realized that focusing on the minute details of a supply chain mattered significantly in achieving practical results.

Vermorel emphasizes the importance of commitment in supply chain management. He contrasts the approach of using fancy technology with the approach of focusing on actual practical results. The latter involves paying close attention to financial risk and performance and spending time understanding the risks and how they manifest in the system. On the other hand, the former approach may involve spending time polishing algorithms, which may not necessarily have a significant impact on supply chain performance.

Throughout the interview, the discussion highlights the importance of supply chain scientists at Lokad, the value of their specialized expertise, and the necessity of focusing on practical results to optimize supply chains effectively.

They discussed the roles and challenges faced by supply chain scientists with Joannes Vermorel, the founder of Lokad, and Maximilian Barth, a Supply Chain Scientist at Lokad. The conversation covers the multifaceted aspects of a supply chain scientist’s role, balancing time between coding, communicating with clients, and handling urgent issues, as well as the importance of avoiding a “zero defect” mentality.

The role of a supply chain scientist, as explained by Barth, involves a technical aspect (coding and understanding client needs), a relational aspect (communicating with clients and identifying the right questions to solve), and a project management aspect (prioritizing tasks and driving projects forward). Vermorel emphasizes that context matters, as pressing issues like a pandemic or an ERP issue may demand immediate attention. Supply chain scientists need to constantly reprioritize tasks based on their potential impact in euros or dollars.

Barth elaborates on the importance of balancing time between implementation and communication with clients. Generally, the balance lies around 20% communication and 80% execution. He points out that it is crucial to strike the right balance between meetings and work to ensure clients’ best interests are met and their expectations are aligned with the work being done.

Vermorel reflects on the challenges of using a classical approach where one person is responsible for client-facing communication and another for the technical side. This method often leads to a loss of information as messages jump between parties. As a result, Lokad adopted a more integrated approach where supply chain scientists handle both the technical and communication aspects, fostering a better understanding of clients’ needs and promoting effective solutions.

Vermorel shares his early experiences in the company, where he wore multiple hats, from sales to data and supply chain scientist. He realized that the conventional method of splitting work among different roles was inefficient and would not scale well.

Vermorel highlights the unique approach taken at Lokad, where supply chain scientists like Maximilian Barth deal directly with clients, eliminating the need for middle managers or software engineers. This approach required dedicated tooling to reduce time spent on technicalities. Barth identifies the key challenge of his role as managing multiple stakeholders while also taking on various responsibilities. He emphasizes the advantage of being the single point of contact, which minimizes knowledge loss during the process.

When asked about transitioning from a data scientist to a supply chain scientist, Vermorel explains that it is actually harder for data scientists than for those with a more general numerical background. He argues that supply chain scientists need to have a taste for quantitative matters, but their focus should be on solving tangible, concrete engineering problems. Data scientists may find it difficult to shift their focus from algorithms to practical solutions, even if those solutions are relatively simple.

Vermorel concludes that while Lokad does employ PhDs, the company’s primary focus is on delivering effective numerical recipes to help clients make high-level, data-driven decisions on their supply chain operations. The discussion revolves around their roles, the company’s diverse workforce, and advice for aspiring supply chain scientists.

Joannes explains that Lokad hires people for software engineering and supply chain problem-solving roles. Data scientists often work on long-term projects, while supply chain scientists like Maximilian focus on solving practical problems with a shorter time frame. Maximilian finds the diversity of problems and the ability to solve them for clients rewarding. He mentions how Lokad’s solutions often automate manual processes, freeing up clients’ time and providing valuable insights.

When asked about the diversity of Lokad’s team, Joannes clarifies that it was not a deliberate choice to create a multicultural team. Instead, the company’s hiring policy is based on skills and competencies, without excluding people based on nationality, language, or gender. He stresses that they prioritize smart, result-oriented candidates, which naturally leads to a diverse workforce.

Maximilian advises those considering a career in supply chain to learn to ask the right questions and see things from multiple perspectives, as projects usually have several stakeholders. Holistic thinking and understanding the needs of all parties involved are essential skills in this field.

Joannes recommends aspiring supply chain scientists to get real-world experience, as opposed to focusing solely on mathematical algorithms or competitions like Kaggle. He emphasizes the importance of understanding the challenges within the actual data, dealing with various stakeholders, and delivering practical solutions that can run without constant supervision. In summary, both guests highlight the importance of hands-on experience, practical problem-solving, and open-mindedness for success in the supply chain industry.

Full Transcript

Kieran Chandler: Previously on this channel, we’ve discussed the importance of a supply chain specialist over someone with more classic data science capabilities. Here at Lokad, this is known as the supply chain scientist, and today we’re lucky enough to be joined by one of our own, Maximilian Barth, who’s going to tell us a little bit more about his daily role and responsibilities. So, Max, thanks very much for joining us today, and perhaps, as always, you could start off by telling us a little bit more about your background and how you came to join Lokad.

Maximilian Barth: Sure. As you can probably tell by my last name, I’m not French, unlike many of my colleagues in the supply chain sciences community. I’m an expat working in France; I’m German. But, like everyone else, I’ve lived in a few different places. I lived in the US growing up, in Finland for a while, and in Australia, and now I’m here in France. Like everyone who’s a supply chain scientist at Lokad, I have a STEM background—science, technology, engineering, and math. What maybe sets me apart a bit is that I have a finance background. I don’t have a classical engineering training or anything like that. However, I actually think finance works well with working in supply chain, as the two are very similar. In finance, you generally optimize your portfolios for returns while minimizing the risk that you could face from markets. It’s very similar in supply chain science. We try to optimize our clients’ inventory for maximum return with minimal exposure to risk and demand variance.

Kieran Chandler: Brilliant. And today, Johannes, we’re talking about a day in the life of a supply chain scientist. I know we’ve spoken about it before, but maybe it’s worth revisiting why you implemented that supply chain scientist capability at Lokad.

Joannes Vermorel: As usual, it was not like a stroke of genius. We tried to have data scientists the traditional way, and it went badly. By the way, I include myself in the first round of inadequate data scientists. Lokad was founded as I was quitting my PhD, which was in computational biology. At the time, that was not exactly the terminology, but it was basically distributed machine learning, so it was data science in all its glory. But it turned out that paying attention to the minute details of a supply chain really matters. Just like Maximilian pointed out, it depends on where your commitment lies. That’s a big question. Does it lie in using fancy tech, or does it lie in getting actual practical results? You might think it’s just a subtle nuance, but actually, the reality entails things that are very different, dramatically different even. I mean, do you care about the financial risk and performance? If so, you’re going to spend time discussing what risks mean, what it means in your system, and how to understand that. Or do you spend time polishing a gradient booster tree so that you can have a slightly more provable convergence proof or whatever that gives you a slightly better algorithm?

Kieran Chandler: Today, we’re going to learn a little bit more about what it is you do in your daily role. What do you see as the core parts of your role?

Maximilian Barth: I think the role is actually very multifaceted. There are multiple components to what you do every day. There’s obviously the technical side – you’re coding a lot, trying to understand your clients’ exact supply chain needs, the nuances of where their challenges lie, and trying to really understand what solution will suit them best and also what exactly their wants and needs are. That also transitions into the next real part of the role, which is the relational aspect of it – being able to talk to clients, trying to figure out the right questions to solve for them, and where their challenges lie. You need to understand what specifically their needs are and what makes them different from someone else, so you can build the right solution for them. I think the third aspect is probably a bit of a project management perspective. We generally sit on projects where, especially if they’re small, we’re the main person driving it at least from the Lokad perspective. So we’re trying to coordinate with our clients, how to best move forward, how to prioritize, and what tasks to tackle first.

Kieran Chandler: What do you see as the most important part of the daily role of a supply chain scientist?

Joannes Vermorel: The most important part really depends on the context. When the supply chain is on fire because of a pandemic or something, first you need to put out the fire. That’s also where, again, the matter of commitment comes into play. If you’re a data scientist, your commitment lies in having a superior algorithm. I believe that frequently, the most urgent and pressing task is much more mundane. The ERP is causing problems because of whatever reason, and the data is just completely out of place. You have duplicate records, you end up with completely incorrect stock records, and so on. Whatever needs to be addressed, it needs to be addressed right now. The problem is that there are so many issues that some of them can potentially be postponed. In terms of resolution, yes, it would be nice if it was 100% clean, but when you’re operating over a sizable supply chain, having zero glitches in anything – the data set you process, the processes themselves, and the way people consume the results you give them – is just not possible. You can’t deliver a zero-defect solution. So at some point, you need to prioritize again based on the financial impact. I believe that in terms of pressure, the supply chain scientist is always kind of reprioritizing what needs to be addressed now, what is important, and what is strategic.

Kieran Chandler: What’s your kind of angle on that, as sort of how do you balance your time between implementing code, communicating with clients, and then how much of your time you spend sort of fighting fires?

Maximilian Barth: I think that’s actually a really good point. Generally, the start of your day is always making sure that there are no fires. You get to the office, check on all of your accounts, and make sure nothing broke overnight. We have clients all over the world in different time zones, so while we sleep, they actually work. Your number one priority is making sure that everything runs as it is supposed to and that our clients have the dashboards ready to use. That was actually my morning today, fixing an ERP change that wasn’t communicated to us. It’s not the most glamorous task, but definitely the most important thing that happened that day. After fixing it, all of the data could be updated. In general, splitting your time depends on the day and week. It’s a tightrope you have to walk. You don’t want to spend too much time in meetings talking to your clients because then you don’t have time to implement anything, but you also don’t want to only work because you may be doing something that’s not in your client’s best interest or what they had in mind. You really have to closely communicate and strike that balance right. I think overall, the balance between actual implementation work and communication with clients, whether it’s through emails or calls, probably lies on average somewhere between 20% to 80%, with 20% communicating and 80% executing what was communicated and discussed.

Kieran Chandler: That conflict’s kind of really interesting, isn’t it? Because you have to spend some more time communicating and having meetings, but obviously you’re spending some of your time feeling like you have to be doing the more technical side of things. It’s a very multifaceted role. Was that always the role you envisioned, or did you ever consider taking a more classical approach where one person would be responsible for the client-facing side of things and another person would be fully focused on the technical side of things?

Joannes Vermorel: We tried the classical way, and the classical way is to have somebody talking to the client, then in turn, this person writes specifications, passes it to IT, and the IT team tries to implement the communication. You end up with a situation where the message goes from one person to another, jumping a few hops, and there’s a very high percentage of information that gets lost in every hop. At the end, you end up with a poor software engineer that implements something that has nothing to do with the problem, with five days of latency just because it had to go through a few people. The problem was that it was kind of broken by design. During the very early years, I could manage, with the help of a few colleagues, to wear one hat where I was the sales guy selling an idea to the client, then start a very dirty implementation, passing the dirty implementation to the software engineer saying, “This thing is working, but in terms of software quality, it’s complete crap. You need to try to make it better, a bit more unit tested, a bit more lean in terms of performance, and maybe a bit more organized.” But they already had the prototype.

Kieran Chandler: The point of this approach is that it’s incredibly archaic in terms of technology, and you need to have somebody who can wear all the hats, you know, from sales to data scientists to supply chain scientists to product manager and everything. So, I realized that this way of splitting the work would never really work well at scale. And by the way, I was saying to our clients at the time that basically everybody was complaining about IT, but IT was complaining about everybody also because the people in IT were saying, “Okay, they say that we do dismal work, but look at the specification and requirement they give us. It’s dismal as well, so you know what, we are on par with them.”

Joannes Vermorel: But that’s just the wrong way to look at it. And basically, what Maximilian does is that there is no middle manager, you know? I mean, you’re literally, and I think that’s something that is pretty unique, the client talks to you. I’m talking about real supply chain practitioners that are literally in the warehouse, facing the stores and everything. And then they talk to you, and you go directly to actually implement the recipe. There is no middleman with a software engineer that you talk to. You don’t coordinate, but in order to do that, we had to engineer some dedicated tooling so that you don’t waste too much time dealing with pure technicalities.

Kieran Chandler: Yeah, I see that as being probably one of the big challenges for a supply chain scientist. You have so many stakeholders and many people pulling on you for your attention, and you’re actually spinning so many plates. It must be pretty difficult. So, what do you kind of see as the key challenge of your role?

Maximilian Barth: I actually think that’s probably the key challenge - that you have to deal with so many stakeholders while also taking on so many different roles. The upside to that is also that you are the one point of contact, and when you’re talking to someone, you’re the person who has discussed the issue with them but also the person who knows what was implemented and how it was done. So there’s not a lot of knowledge that gets lost in translation between the multiple steps. I think that’s the main advantage, but also obviously the main challenge because you have to be able to do so many different things quite well. You want to be able to actually write a good solution for your clients but also really understand what it is that they need.

Kieran Chandler: Yes, definitely. There’s no running away from it, is there? If you’re in trouble and you’ve done something wrong, you’re definitely held accountable. For a supply chain scientist, we’re very adamant they need to have some supply chain expertise. If you were a data scientist, how easy would it be to transition into the role of someone like a supply chain scientist like Max?

Joannes Vermorel: The paradox is, and I believe that’s the case, it’s actually way harder for data scientists to transition to supply chain scientists than it is for generic engineers or people that are numerically minded in general. It’s funny, there are two words in French that exist, but I don’t think there’s an English translation. It’s the difference between a mathematician and a “matheux,” you know, someone who’s mathematically inclined.

Kieran Chandler: So, the first question I have is about the qualities required to become a supply chain scientist. Joannes, can you tell us what kind of people you’re looking for?

Joannes Vermorel: What we need is people that have a taste for numbers. Supply chains are large, and you can’t just have an intuition of thousands of products. You need to have people that have a taste for quantitative matter in general. But the trick is that it’s a very applied role. Maximilian is literally helping companies to make decisions on millions of euros of physical assets. It’s really tangible decisions that get made at the end of the day. You need to have this mindset that is a very concrete engineering problem that you’re addressing. And I know that it might not please the data science audience, but my experience was that it’s actually very difficult to convert people who have been doing data science for a few years to actually become good at what we call a supply chain scientist because, again, the focus is not the algorithm. The focus is the fact that you have a numerical recipe. Another episode that we recently did was that the numerical recipe end-to-end really makes sense at a very high level, and it doesn’t really matter whether it’s sophisticated or not. If you can basically get away with a semi-trivial solution, excellent job done and just adjust, then simply, yes, you won’t get a paper just because you figured out that a slight numerical coefficient adjusted just the right way does the magic. You can’t publish on that. But if it does the job, you know, why not?

Maximilian Barth: Yeah, and so our experience was that although we do have PhDs that do that, I mean, we have people doing PhDs at Lokad actually. We have five in total. Two have already completed their PhDs, and three are still in progress, but I’m confident that they will be able to defend their PhDs. But literally for us, those people are on the pure software engineering tracks, where it’s not even the same timeline. I mean, people are attacking problems and thinking of delivering a solution over the next three years. That’s the timeline of a data scientist on the platform side, so they think about something like differentiable programming. We have somebody doing a PhD on differentiable programming, and this person is literally building and engineering the building blocks of differentiable programming, but this person is not solving any actual supply chain problem. Maximilian is doing that. And when you’re working, I mean, I’m not too sure on which problem you are, but typically the time frame is, you’re looking from one day ahead to maybe a couple of months ahead, but certainly not like three years ahead. It’s just not even the same time scale.

Kieran Chandler: And what about the rewarding part of the job, what do you really enjoy?

Maximilian Barth: I think the most rewarding thing is probably the diversity of problems you face and then how tangibly you can actually solve them. A client comes to you with something that they’ve found no solution for yet or only a really bad manual work around where they work a lot in Excel, and then we can deliver something that automates that.

Kieran Chandler: It seems like there’s a lot of value Lokad can deliver for clients, not just in terms of decision-making, but also in providing a clearer picture of their business. Can you talk a bit about the multicultural and international aspects of the Lokad team, particularly the supply chain scientists? Why did you create a team like that and why was it important to you?

Joannes Vermorel: First, I want to clarify that our team’s diversity wasn’t a specific intent, but rather a result of not excluding certain classes of people. In other words, we didn’t create a diverse hiring policy to deliberately choose people from specific countries or backgrounds. My thinking was that I didn’t want Lokad to have any aspect in its design that would exclude people.

For example, we have about one-third of women in the company, which is quite a lot for a tech company. To achieve that, you need to ensure that the company’s setup isn’t directly adverse to young women or promotes behavior that would make them feel unwelcome.

Similarly, we are based in Paris, but if we required perfect French, it would narrow down our options to mostly French people or those from former French colonies. By removing such barriers, we attract a diverse pool of candidates who we can then evaluate based on their skills.

It turns out that French people don’t have a monopoly on intelligence and diligence. So now, only about 40% of our employees are French nationals, with the rest being mostly from the European Union and beyond. This diversity isn’t a result of active discrimination, but rather a focus on judging people based on their abilities to be smart and get things done.

Maximilian Barth: That’s right, and it’s important to emphasize that by removing these barriers and focusing on skills, we’re able to bring together a team with diverse perspectives and experiences. This ultimately benefits the company and our clients, as it allows us to approach problems from multiple angles and find more innovative solutions.

Kieran Chandler: Well, it turns out that frequently we end up with non-French people being hired. I mean, I really love France, that’s my country where I’m born, but as an employer, I need to hire first and foremost the people who will serve the company best, not just the ones that ended up being born in the right place.

Maximilian Barth: It’s a good job you didn’t rule out perfect French because I think both me and Max would have been in a world of trouble.

Kieran Chandler: If you start wrapping things up a little bit now, Max, what kind of advice would you give to somebody who’s maybe considering a career in the supply chain industry or indeed as a supply chain scientist?

Maximilian Barth: I think for my job, the most important skill that I find the most valuable, and what’s helped me the most, is to learn to ask the right questions and to see things from multiple perspectives. Our projects generally have multiple stakeholders, and while you may have one contact person who you talk the most with, you always have to remember that they also represent other people. So you want to really make sure that you find a holistic solution that fits everyone well. I think being able to see things from a holistic perspective and being able to ask the right questions is probably the most important skill you want to develop.

Kieran Chandler: Okay, great. And Joannes, to conclude, what advice would you give to someone who’s maybe an aspiring supply chain scientist?

Joannes Vermorel: Get real. There are so many ways to be not real. Kaggle is fantastic, but it’s just games. Algorithms are fantastically interesting, but it’s not real, at least not directly. My suggestion is if you want to start a career in supply chain, you need to get your hands dirty. Delve into an ERP, start looking at what the data really looks like, not the idealized version that you get in textbooks that is already completely cleaned and perfect and well-arranged. And indeed, deal with having many shareholders. That’s really difficult because how good is your solution if the company ends up fighting among itself? You have to find a way so that the solution is acceptable to all those various parties. That’s a very tough challenge, but you have to do it while preserving your engineering values. You want to have something akin to a capitalistic process. You’re not just a consultant that produces PowerPoints and delivers things. Maximilian, you are delivering something that is working in production and hopefully can even run without you, so there is a real asset that is being engineered and improved over time. It’s not PowerPoints that are being delivered.

Kieran Chandler: Okay, brilliant. We’ll have to wrap it up there, but thanks both for your time.

Maximilian Barth: Thank you.

Kieran Chandler: That’s everything for this week. Thanks very much for tuning in, and we’ll see you again in the next episode. Thanks for watching.