Data Scientist- Process Modeling & Machine Learning

SMS group Inc•Pittsburgh, PA

2d•Hybrid

About The Position

We are seeking a Data Scientist who combines strong machine learning expertise with a genuine understanding of the processes behind the data. In this role, you will collaborate closely with process engineers and domain experts to understand how our machinery and production processes behave and apply that knowledge to develop models. The central focus of the position is enhancing our existing process models with data-driven techniques and machine learning. By integrating domain knowledge with advanced modeling methods, you will build solutions that perform robustly in production and earn the confidence of the engineers, operators, and customers who depend on them.

Requirements

Master’s degree in data science, Machine Learning, Statistics, Applied Mathematics, a quantitative engineering discipline (e.g., process modeling, mechanical, control), or a related field - or 2+ years of relevant experience.
Proven experience taking AI/ML solutions into real production environments (not just notebooks and prototypes).
Demonstrated ability to understand a problem domain and incorporate that understanding into models - comfort working alongside engineers and domain experts and learning the underlying process.
Understanding of software development practices: Python, SQL, Git, code review, and familiarity with container technologies.
Strong communication skills for working with other departments, customers, and stakeholders, and the ability to explain technical choices to non-specialists.
Ability to plan over longer horizons and coordinate work packages effectively.
Willingness to work on-site at the office and to travel to customer sites.

Nice To Haves

Hands-on experience with hybrid / gray-box / physics-informed modeling, or with enhancing first-principles or simulation models using data-driven methods.
Experience building virtual/soft sensors, digital twins, or model-based monitoring for industrial or physical processes.
Proficiency with deep learning methods and frameworks.
Background in or exposure to an industrial / manufacturing / process domain (steel, metals, chemical, energy, or similar).
Experience researching and benchmarking existing solutions and algorithms before building from scratch.

Responsibilities

Partner closely with process engineers, metallurgists, and domain experts to develop a deep, working understanding of the underlying processes and the data they generate.
Translate process know-how into model structure — constraints, features, and relationships — rather than treating the process as a black box.
Spend time where the data comes from: participate in site visits to connect raw signals to real physical behavior.
Enhance existing process models with data-driven techniques, replacing weak assumptions or unmodeled effects with learned components while preserving the physics that already works.
Design and implement hybrid models that combine domain/first-principles sub-models with machine learning (gray-box, physics-informed, and residual-modeling approaches).
Develop virtual/soft sensors to estimate quantities that are hard or expensive to measure directly.
Own data science problems end to end: scoping, data analysis (time-series and relational), feature engineering, modeling, validation, and deployment into production.
Serve as the algorithmic point of contact for your solutions, choosing the right tool for the problem — robust feature-based methods (e.g., scikit-learn) as well as deep learning (e.g., TensorFlow/Keras) where it adds value.
Build engineering prototypes and turn promising experiments into reliable, maintainable production solutions.
Monitor model performance on live data, diagnose drift, and retrain or recalibrate as conditions change.
Continuously optimize and maintain deployed solutions to improve accuracy, robustness, and runtime performance.
Work with cross-functional teams (product, engineering, project management) to align data science work with product roadmaps and project goals.
Engage with customers to gather feedback, refine solutions, and ensure that data-driven approaches are accepted and adopted by the people who use them.