Senior, Data Scientist / Research Engineer
Role details
Job location
Tech stack
Job description
At Schneider Electric, we are committed to solving real-world problems to create a sustainable, digitized, new electric future. Artificial Intelligence has the potential to transform industries and help unlock efficiency and sustainability. Within our Global AI Hub we combine our long-standing manufacturing and domain expertise with cutting-edge innovation in AI, machine learning, and deep learning to empower smarter decision-making, agility, and decarbonization. Our Strategy & Innovation team drives the AI strategy and innovation efforts for the AI Hub, Schneider Digital, and Schneider Electric at large. We are building the next generation of intelligent systems that combine large-scale multimodal models, graph/topology understanding, simulation, and domain logic to enable system-level reasoning across energy, buildings, industry, and data centers. Your role : We're looking for a curious, fast-moving applied research scientist who loves turning cutting-edge papers into prototypes and pushing them further. You combine strong transformer fundamentals with rigorous experimentation, solid engineering habits, and an end-to-end maker mindset - from preparing the data to building the model to crafting demos that make the value visible. You thrive in a collaborative environment, engage actively with the research community, and enjoy working with product and business teams to translate ideas into real impact.
Your responsibilities :
- Develop advanced representations for core modalities - time series, tabular, text, and graph/topology, visual/3D data
- Rapidly translate state-of-the-art research into prototypes, adapting multimodal and transformer-based architectures to Schneider-specific datasets
- Build robust, reproducible ML pipelines, covering data preparation, experiment tracking, baselines, ablations
- Lead the creation and preparation of multimodal datasets, transforming raw data (such as time-series signals, structured tables, documents, diagrams, and system relationships) into clean, usable training datasets
- Collaborate with domain experts and product teams to align modeling choices with physical constraints and convert prototypes into clear, impactful demonstrations, * The opportunity to shape the next generation of multimodal AI systems for energy, industry and sustainability
- Access to rich, real-domain multimodal datasets, rarely available in academic or tech environments
- A role at the intersection of AI research, physical systems understanding and sustainability, working on problems with real impact
- A fast moving, collaborative, and deeply technical team, embedded within Schneider Electric's global AI strategy and innovation ecosystem
Requirements
Do you have a Master's degree?, * PhD in Machine Learning, AI, Computer Vision, NLP, Robotics, or related field with strong foundations in transformers and modern representation learning. We are also open to exceptional candidates from top Master's programs or leading engineering schools with demonstrated research or applied excellence
- Demonstrated experience in multimodal ML, with strong skills in at least two core modalities:
- Time-series modeling (e.g., forecasting, anomaly detection, foundation models)
- Tabular/structured data modeling or feature representation
- Text understanding (LLMs, embeddings, document understanding)
- Graph learning (GNNs, message passing, topology-aware modeling)
- Visual modalities (2D/3D)
- Strong hands-on experience with PyTorch, custom model architectures, and efficient training/finetuning methods
- Ability to design clean, rigorous experiments (baselines, ablations, evaluation protocols) and communicate findings clearly
- Solid engineering discipline: Git, PRs, code reviews, reproducibility, experiment tracking, and collaborative development practices
Preferred Skills
- Interest or familiarity with engineering, energy, or physical systems - curiosity about real-world technical domains is a strong plus
- Exposure to simulation-based learning, physics-aware models, or neuro-symbolic approaches
- Comfortable moving between research and applied prototyping, turning ideas into working demos
- Contributions to open-source projects, workshops, or scientific publications