Founding Chief Scientist – Physics AI & Foundation Models

We're not recruiting employees.We're assembling an exceptional founding team.Neodustria is preparing the launch of a Swiss deep-tech program headquartered in the Canton of Jura, with a clear ambition:To build Europe's next-generation Industrial Intelligence Platform.We believe the future of manufacturing will not be driven by isolated AI tools, disconnected engineering software or fragmented industrial data.It will be built through the convergence of Artificial Intelligence, Engineering, Physics Simulation, Industrial Knowledge, Market Intelligence and Human Expertise into a single intelligent platform helping manufacturers design, simulate, decide and operate more intelligently.Over the coming months, we will assemble a small, multidisciplinary founding team of exceptional builders.This is not a traditional recruitment campaign.There are no predefined executive positions waiting to be filled.There are no shortcuts to leadership.Instead, we are looking for extraordinary individuals who want to help create one of Europe's most ambitious industrial deep-tech companies from the ground up.We are not looking for the best CVs.We are looking for the people who will build the company others will study ten years from now.------------------FOUNDER TRACKThe Founder Track is designed for exceptional individuals who wish to participate in Neodustria's development phase.It is not a traditional employment position.It is a structured collaboration period during which both parties evaluate:Technical excellenceOwnership mindsetScientific rigorLeadership potentialCultural fitLong-term alignmentShared ambitionWe recognize that building a deep-tech company requires conviction from everyone involved.We are therefore looking for individuals who are prepared to invest their expertise before expecting immediate rewards because they believe in creating long-term value rather than pursuing short-term compensation.ABOUT USNeodustria is building a sovereign Industrial Intelligence Platform combining engineering, physics, geometry, artificial intelligence and industrial knowledge.Our ambition is not to build another simulation interface or generic AI assistant. We are developing proprietary scientific intelligence capable of learning from CAD models, meshes, materials, boundary conditions, simulation results and real industrial data.The platform integrates physics solvers such as OpenFOAM and CalculiX with scientific machine learning, knowledge graphs, geometry intelligence and a high-performance runtime designed for industrial deployment.THE MISSIONWe are looking for a rare scientific and systems builder to lead the development of the Neodustria Physics Foundation Model.You will create a proprietary model capable of learning representations of geometry, materials, physical conditions and engineering behaviour, then producing reliable physical predictions with efficient CPU inference.This is not a prompt-engineering, RAG or generic MLOps position.You will own the full scientific intelligence lifecycle:Simulation data → physical representation → model architecture → training → validation → compression → Rust runtime → industrial deploymentWHAT YOU WILL BUILD Your first responsibility will be to architect and deliver the Neodustria Scientific Intelligence Runtime, including:A canonical representation for CAD, meshes, materials, loads and boundary conditionsSimulation-data pipelines connected to OpenFOAM and CalculiXGeometry, mesh, material and physics encodersSurrogate and foundation-model architectures for scientific predictionFull-field prediction of stress, displacement, temperature, pressure and velocityUncertainty quantification and confidence estimationPhysics-aware validation against numerical solvers and experimental dataModel compression, quantization and distillationA high-performance Rust inference runtime optimized for CPUsAPIs and SDKs for integration into the Neodustria platformScientific benchmarks, patentable technologies and publishable researchREQUIRED SCIENTIFIC EXPERIENCEYou should have strong expertise in several of the following areas:Scientific machine learningPhysics-informed learningNeural operators such as FNO or DeepONetGraph neural networks and mesh-based learningReduced-order modellingSurrogate modellingDifferentiable simulationBayesian inference and uncertainty quantificationActive learning and multi-fidelity learningInverse problems and optimizationContinuum mechanics, CFD, FEA or multiphysicsNumerical methods, sparse linear algebra and iterative solversThree-dimensional geometry and scientific data representationsYou must understand the limitations of both machine learning and numerical simulation and know when classical physics should remain the primary computational method.REQUIRED TECHNICAL EXPERIENCEYou must be able to transform research into a production-grade industrial runtime.Strong experience is expected in:Python and PyTorch or JAXRust for high-performance production systemsCPU inference optimizationMultithreading and parallel computationSIMD and vectorizationMemory layout, cache efficiency and zero-copy processingQuantization and model compressionONNX, MLIR, TVM or related compiler/runtime technologiesC or C++ interoperabilityProfiling, benchmarking and numerical validationDistributed training and HPC environmentsExperience with OpenFOAM, CalculiX, PETSc, Gmsh, OpenCascade or comparable engineering technologies is highly valuable.THE CANDIDATE WE ARE LOKKING FORWe are not looking for someone who simply selects an existing neural-network architecture.The right person will think systematically about:how industrial physics data should be representedhow models generalize across geometries and mesheshow conservation laws and physical constraints enter training and validationhow uncertainty and numerical error are communicated to engineershow inference runs efficiently without dependence on expensive GPU infrastructurehow every industrial program creates reusable models, datasets and intellectual propertyYou should have evidence that you have built - not merely studied - scientific models, numerical systems, compilers, runtimes or production inference infrastructure.YOUR POSITIONYou will work directly with:PHDs in Artificial IntelligenceThe Physics Intelligence teamThe Geometry Intelligence teamThe Knowledge and Ontology teamThe Platform Systems teamIndustrial transformation and customer-program teamsYou will have the authority to define the architecture, research roadmap, technical standards and initial team required for Neodustria’s proprietary scientific intelligence technology.WHY JOINThis is an opportunity to create a foundational technology rather than optimize an existing product.You will help build a model that understands the language of industrial engineering: geometry, materials, forces, flow, heat, deformation and failure.