Sadush Zeqiri: Data Science at wienerberger

Since 2022, Sadush Zeqiri and her team have been tackling a truly massive challenge: extracting actionable insights from large datasets. The aim is data-driven production optimization. Through data-based decision-making, production is becoming smarter, more efficient, and future-proof.

On the Way to the Smart Factory

The transformation is well underway – and its potential is enormous. “Our job is to drive digital transformation at wienerberger. We pursue three main goals: reducing downtimes, improving product quality, and saving materials,” explains Sadush Zeqiri. This can be illustrated with three concrete examples:

• Avoiding downtime: By analyzing collected data, operations teams can identify patterns and causes of failures, enabling them to intervene quickly and prevent prolonged interruptions in production.
• Reducing waste: Data automatically generates relevant KPIs, helping to detect quality deviations early. This allows for timely countermeasures before more material is wasted.
• Lowering resource consumption: In the piping production of the Pipelife brand, so-called mass-balance dashboards help monitor material usage. They flag when too much or too little material is being used. Both are problematic: one increases costs, the other jeopardizes product quality. With these dashboards, material usage at the Pipelife plant has been significantly reduced without compromising quality.

Data Science for Greater Sustainability and cost reduction

When asked about the greatest potential of her work as a Data Scientist at wienerberger, Sadush Zeqiri’s answer is clear: energy optimization through data science. This is also a key lever in advancing the company’s sustainability goals. Analyses of material and energy consumption not only provide valuable insights but also form the basis for saving resources and significantly reducing CO₂ emissions.

“Data shows us how we can optimize material usage, reduce energy consumption in production, and consequently cut CO₂ emissions,” says Sadush Zeqiri.

A first use case has already been implemented: at a plant in Poland, an AI model calculates the conditions under which products can be fired with the lowest possible energy while maintaining high quality. This improves energy efficiency in the kiln process and reduces CO₂ emissions.

Small Team, Big Challenge

Before data can create value, it must first be collected, processed, and intelligently linked on our central data platform, namely Databricks.

This is a challenge for the Data Science team, which consists of only three people, including Sadush Zeqiri. Reality is far from homogeneous.

“We collect data from more than 50 sites. They come from various sources, in different formats, often with inconsistent naming,” explains the expert. The team faces three major challenges:

1. Decentralized data sources: Information is stored locally in different systems. “It is a major challenge to consolidate them centrally,” says Sadush Zeqiri.
2. Inconsistent naming: “In the various countries where we operate, sensor values often have different names. This makes standardization difficult,” she adds.
3. Complex integration of large data volumes: With millions of data points to process, the small team of three quickly hits its limits.

But which data does Sadush Zeqiri actually need for her work?

“For the current use cases, we rely on data from the Manufacturing Execution System (MES). It enables us to collect structured, standardized information, visualize production plans, and provide context for production activities. This includes recording shift details, orders, downtimes, output volumes, and quality metrics. We also collect data from machine sensors, such as temperature readings, conveyor belt speeds and a wide range of other operational parameters.” 

Problem to Use Case

Before starting any project, Sadush Zeqiri asks three key questions:

1. Business impact: What concrete benefit will the project bring?
2. Availability: Are the necessary data even available and accessible?
3. Strategic relevance: Does the initiative align with wienerberger’s strategic goals?

These questions help turn ideas into viable use cases. The Data Science team works closely with the production operations teams, as their experience and practical knowledge are crucial for developing meaningful, feasible applications.

"Exchange with production staff is essential. They know the machines, daily challenges, and practical constraints,” says Sadush Zeqiri. “For every use case, we start with a pilot project. This allows us to quickly see whether the idea works – and the real value it delivers.”

Significant Impact on Daily Operations

“Data is only valuable if it is understood and used,” emphasizes Sadush Zeqiri. She also acts as a bridge-builder between abstract data analysis and operational requirements.

Especially at the beginning of her time at wienerberger, she trained employees to use dashboards and taught them how to extract relevant data for decision-making.

Today, data-driven tools are an integral part of everyday operations at wienerberger. With just one click, employees can retrieve parameters such as line speeds, machine run times, or rejection rates over time and use them to adjust production processes.

“Teams mainly use the data in the daily morning meetings to review how production performed the previous day. Data brings more transparency to processes and helps people make fast, informed decisions,” says Sadush Zeqiri.

Machine Learning in Manufacturing

Sadush Zeqiri’s work also benefits from technology partnerships, such as with data analytics company SAS. “They help us implement data science in production,” she explains. SAS has supported wienerberger with machine learning for energy optimization in kiln processes.

More use cases – such as intelligent monitoring systems – are in the pipeline. In the future, wienerberger plans to increasingly use machine learning models in production.

Quality Prediction and Control

By leveraging real-time analysis of process parameters, product quality can be forecasted during production, allowing deviations to be detected early. “This enables us to take corrective actions immediately, reduce scrap, and ensure consistent product quality,” explains Sadush Zeqiri.

Optimizing Production Planning

The third area with great future potential is data-driven optimization of production planning. By combining and analyzing data from order management, shift planning, maintenance, and warehouse logistics, orders can be better planned, maintenance needs anticipated, and material shortages avoided.

This ultimately improves delivery reliability. “The next stage would be to automate all of this,” Sadush Zeqiri says, looking ahead.

With this clear future agenda, Sadush Zeqiri is actively shaping wienerberger’s digital transformation. Her expertise and commitment make her a driving force in building a more efficient and sustainable industry. Data science is not just about analysis – it is the key to a new industrial reality.

Learn more about the work of successful women at wienerberger in our blog post Breaking Barriers: Successful Women in Male-Dominated Fields at wienerberger.