AI and data analytics are game changers for the manufacturing industry’s demand and shop floor agility. My team is currently developing an AI model that identifies irregular demand patterns and updates itself.
Driving Supply Chain Innovation Through AI and Process Optimization
Q&A with Juliet Mirambo, Operations Leadership Development Program | MilliporeSigma
You’re part of MilliporeSigma’s Operations Leadership Development Program, which rotates future leaders through core manufacturing and supply chain functions. How has this experience shaped your understanding of end-to-end manufacturing operations at a global scale?
From the material selection to the design of the configuration, all things impact manufacturing lead-time and delivering to customers. Simply said, forecasting is not only a planning exercise but a vehicle to accelerate the manufacturing pace. In MilliporeSigma, I experienced this while working on the process of handing over the demand for configurable materials to manufacturing. As there are many configurations for this type of products, it was easy to see how any deviation between forecasted and actual requirements for the shop floor translated directly to a manufacturing delay. We worked across commercial, demand planning, and operations teams on developing a process that was more standardized and resilient to changes.
This project helped me to understand the importance of bringing supply chain concepts early on to the design and planning process. The understanding of these concepts is even more crucial at a global level due to the many factors affecting international supply chains such as tariffs and customs and border delays which affect service level and delivery time. In order to manage this complexities and drive value to supply chain efficiency we must look at the solutions from an Industry 4.0 perspective, with new technologies such as AI, IoT, and digital twins that help to integrate systems.
As a process optimization lead, you’re working within a €250 million planning core model. What are the biggest challenges in optimizing material flow and demand planning for configurable manufacturing components?
My exposure with a €250 million Planning Core Model places me in context of some of the most relevant and ongoing problems concerning optimization of material flow and demand planning for configurable components used in manufacturing. Out of the multiple areas, I see lack of end-to-end supply chain visibility and fragmented data as a potential source of complexities. Using my project at MilliporeSigma as an example, trying to find packaging locations of a product can become a rather tedious process all due to fragmented master data. Key data elements can be scattered over multiple systems, and when not fully synchronized, can lead to planning errors and delays. Another example, purchase order details may exist in SAP ERP, but messages about delays, exceptions are often discussed in email. These are well-known blind spots plaguing the planning process and this is why our daily focus has been placed on advocating a common data structure that can facilitate real-time visibility.
AI and data analytics are transforming manufacturing. How are you currently exploring AI to improve forecast accuracy and operational agility on the shop floor or in planning?
AI and data analytics are game changers for the manufacturing industry’s demand and shop floor agility. My team is currently developing an AI model that identifies irregular demand patterns and updates itself. Some demand is stable and seasonal while other items show demand once every few years. Forecasting based on historical averages (or human forecasting) is not good enough. This is why we are using AI-based forecasting based on Python and KNIME, so the model can capture subtler signals or track changes in customer demand. Next, we are also using Minitab to quantify prediction accuracy and its variability. This means differences over time can be normalized and we can ‘prove’ improvements and model performance. AI also helps automate conversions of units of measure of configurable materials. Sounds all technical and specific, but really if the company doesn’t have a repeatable base unit and isn’t automatically reconciled…then all orders are ‘off’. Reconciling units of measures in the forecast means cleaner data to work with – a must have or starting point for any AI model.
With your background in engineering and chemistry, how do you bridge the gap between technical product knowledge and operational efficiency in manufacturing environments?
My dual education in engineering and chemistry gives me a unique opportunity to facilitate the cross-functionality between in-depth technical product understanding and its efficient manufacturing execution. My first rotation at MilliporeSigma was in materials management dealing with several hazardous materials. My chemistry knowledge assisted me in understanding the hazard classification processes and environmental health and safety requirements. Technical knowledge is also critical in a biotechnology company like MilliporeSigma as most of the products are highly temperature-sensitive and need precise supply chain and cold chain logistics. My work experience at MilliporeSigma is further supported by my engineering understanding of continuous improvement of processes. Whether it is to enhance a process for waste or cost reduction or to enable a technology transfer, I cultivate an improvement mindset. For example, I am currently involved in the migration to a new forecasting software which greatly enhances the efficacy of demand forecasting. From this experience, I am learning the real importance of aligning the technical understanding with the functional work streams.
End-to-end process mapping is central to your role. What have you learned from mapping complex manufacturing and supply chain workflows that surprised you or shifted your thinking?
I completely agree! End-to-end process mapping has been a huge revelation for me in this role at MilliporeSigma. One of the projects that has significantly influenced my perspective, is the ongoing effort to improve the demand planning handover process to operations. The aim is to make the forecast signal sent to production more accurate and usable. This seems like a simple task, but so far, it has started to uncover a lot of the implicit assumptions and gaps that existed. One example is the adjustment of forecast based on market “tell” conversations. When an adjustment is made, if it is either not documented, or not communicated downstream to other teams, the teams need to start speculation. Did the demand planner receive additional information which caused a change? The lack of definitive information will result in swings away from targeted levels and a ripple effect of inefficient planning across the system.
Given the ongoing volatility in global supply chains, what strategies do you believe manufacturing leaders need to prioritize to build resilience and responsiveness?
Too many organizations still remain too reactive when facing disruption. Instead of leading their workforce in recovery, manufacturing leaders must invest in predictive systems that serve as early warning and visibility into risks. For instance, advanced tools that rely on AI and machine learning are now able to analyze historical and current data to detect anomalies with suppliers, weather, geopolitical conditions, logistics and more. One of the systems I am working on now which works towards this goal is called Supply Pulse. The purpose of this tool is to consolidate and centralize our purchase order system data to capture metrics around delivery performance, and then notify supply chain managers of shipments that are already known to be at risk or late delivery, ahead of their downstream impact. Meanwhile, as much as resilience is about what you do produce, it’s just as much about what you don’t. This is where the lean operating model comes into the picture. Eliminating over production can significantly reduce waste while creating a more agile and flexible network .
You volunteer with the SPARK™ program and lead hands-on science experiments through the Curiosity Cube. How does your experience in manufacturing inform the way you mentor students and promote STEM education?
Being in the industry of manufacturing and supply chain optimization, I know how important curiosity and problem solving is to drive new innovations on the shop floor and it’s that same curiosity and problem identification that we want today’s students to have. When I engage young learners, I connect the dots between the science experiments they perform at the Curiosity Cube and real-life applications. When we examine reactions and temperature, for example, I discuss how crucial it is to monitor temperature levels while delivering vaccines.
What advice would you give to early-career professionals, particularly women, who are looking to enter or grow within manufacturing, operations, or industrial supply chain roles?
Young professionals should know that they are capable of doing more than they think they can. I remember when I started my first rotation in materials management and I spent nine months on four different simultaneous initiatives—reducing waste in the warehouse, finding a new 3PL, inventory-reconciling missing materials, and moving product to a new warehouse. It was a lot for a new professional. At the end of that initiative, a contract was signed and allowed customers to get products from our manufacturing facility much sooner.
Finally never forget the big picture of any problem. Who are the players? What are the implications of not solving this? How is my design solution going to affect them? With the big picture in mind, you can develop more target solutions that are in harmony with the expectations of the consumers. You'll also be spared from overlooking critical aspects that may result in greater costs down the line.
The content & opinions in this article are the author’s and do not necessarily represent the views of ManufacturingTomorrow
Comments (0)
This post does not have any comments. Be the first to leave a comment below.
Featured Product
