TVARIT GmbH specializes in developing and delivering cutting-edge artificial intelligence (AI) solutions for the metal industry, including steel, aluminum, copper, cast iron, and more. Our software products empower customers to make intelligent, data-driven decisions, driving

advancements in Predictive Quality (PsQ), Predictive Maintenance (PdM), and Energy Consumption Reduction (PsE).

With a strong portfolio of renowned reference customers, state-of-the-art technology, a talented research team from prestigious universities, and recognition through esteemed awards such as the EU Horizon 2020 AI Prize, TVARIT is recognized as one of the most

innovative AI companies in Germany and Europe.

We are looking for a self-motivated person with a positive “can-do” attitude and excellent oral and written communication skills in English.

We are looking for a Senior Data Scientist to join TVARIT’s Metals Division and build AI solutions that improve quality, yield, equipment reliability, and energy efficiency in metal manufacturing plants. The role suits a hands-on data scientist who has worked with real plant data, is fluent in time-series analysis and its tooling, and brings expertise in at least one metal manufacturing domain.

You will work directly with process data from customer plants, collaborate with process engineers and plant teams to understand the physics behind the data, and take models from exploration through validation and deployment. You will also contribute to reusable models and components that scale across plants and customers.

• Develop, validate, and improve ML and statistical models for Predictive Quality, Predictive Maintenance, and energy optimization use cases in metal production and processing.

• Analyze high-frequency and batch time-series data from sensors, PLC/SCADA systems, historians, MES, and quality labs, including alignment of process signals with quality and batch or coil/heat records.

• Engineer features that capture process behavior, such as lags, rolling statistics, frequency-domain features, process-stage segmentation, and physically meaningful derived variables.

• Apply anomaly detection, forecasting, classification, regression, and root-cause analysis techniques, and combine data-driven models with process knowledge and physics-based constraints where appropriate.

• Design robust validation strategies for time-dependent data (e.g., time-based splits, leakage checks, drift analysis) and quantify model uncertainty and limitations clearly.

• Work with plant process, quality, and maintenance engineers to understand process flows, operating conditions, and failure modes, and translate them into well-defined modeling problems.

• Assess data readiness and quality at customer sites, identify gaps in instrumentation or data capture, and recommend practical remedies.

• Present findings, model results, and recommendations clearly to plant teams and customer stakeholders, linking insights to measurable outcomes such as scrap, yield, downtime, and energy consumption.

• Contribute to end-to-end project delivery, from data onboarding and exploratory analysis through model deployment, monitoring, and retraining, working closely with software and data engineering teams.

• Write clean, tested, reusable Python code and contribute to shared libraries, pipelines, and model templates that can be configured across plants and customers.

• Mentor junior data scientists, review their analyses and modeling approaches, and help establish best practices for industrial data science within the division.

• Bachelor’s or master’s degree in Data Science, Computer Science, Statistics, Mathematics, Metallurgical, Mechanical, Electrical, or Chemical Engineering, or a related discipline; a PhD is an advantage.

• 7+ years of overall professional experience, with substantial hands-on experience in data science or machine learning applied to industrial or manufacturing data.

• Mandatory manufacturing plant experience: has worked on data science projects with manufacturing plants, using real process, sensor, and quality data, and working directly with plant operations or engineering teams.

• Mandatory time-series expertise: strong practical experience in time-series data analysis, including preprocessing, resampling, alignment, feature engineering, anomaly detection, forecasting, and validation of time-dependent models.

• Metal manufacturing domain expertise: in-depth knowledge of at least one metal manufacturing domain, such as steelmaking (BOF/EAF, secondary metallurgy, continuous casting), aluminum (casting, rolling, extrusion), copper, foundry/die casting, forging, rolling, or heat treatment, including the key process variables and quality drivers.

• Proven track record of taking models beyond notebooks to validated, deployed solutions that delivered measurable business impact.

• Strong written and verbal communication skills, with the ability to explain technical results to both engineers and non-technical stakeholders.

• Willingness to travel to customer plants as required.

• Programming: Strong Python skills for data science, with working proficiency in SQL; good coding practices including version control (Git), modular code, and testing.

• Data analysis libraries: pandas, NumPy, SciPy, and visualization libraries such as Matplotlib, Plotly, or Seaborn.

• Time-series and ML packages: Hands-on experience with packages such as statsmodels, scikit-learn, sktime, tsfresh, Darts, or Prophet, and gradient boosting libraries such as XGBoost or LightGBM.

• Deep learning: Experience with PyTorch or TensorFlow for sequence models (e.g., LSTM, temporal CNN, transformer-based models) is an advantage.

• Statistics and modeling: Solid foundation in statistics, signal processing, regression, classification, anomaly detection, forecasting, and design of

experiments; familiarity with hybrid physics-informed modeling is an advantage.

• Industrial data: Familiarity with PLC/SCADA data, historians (e.g., OSIsoft PI or similar), MES, and lab/quality systems, including data quality and contextualization challenges.

• Deployment and MLOps: Understanding of model deployment, monitoring, drift detection, and retraining; experience with tools such as MLflow, Docker, and cloud platforms (Azure preferred; AWS or GCP also relevant) is an advantage