Assemble compute for materials science.

Software infrastructure for orchestrating materials research pipelines.

The pipeline

A graph of computational work.

A Tensir pipeline is a graph whose nodes represent computations or transformations: structure ingest, density-functional theory (DFT), machine-learned interatomic potentials (MLIP), molecular dynamics, relaxation, scoring, ranking, or refinement.

Tensir provides one interface for managing these heterogeneous computations across a research pipeline. Execution can span public cloud, HPC, and on-premises infrastructure.

Product

Five responsibilities. One system.

Pipeline

The graph of computational tasks and transformations. Nodes describe the work; connections describe how that work fits together.

Compute

Allocation and execution of workloads across public cloud, HPC, and on-premises infrastructure.

Budget

Cost estimation, spend control, and recorded authorization for computational work.

Provenance

Execution context stays attached to results: source, method, parameters, licence class, operator, cost, and other context from the run.

Tree

A hierarchical planning and reasoning layer over the pipeline. It can support bounded LLM or agent tasks: suggesting nodes, restructuring parts of a pipeline, optimizing cost, or proposing next computations.

Methods and materials

Computations and objects of work.

Scientific methods

  • Density-functional theory (DFT)
  • Machine-learned interatomic potentials (MLIP)
  • Molecular dynamics

Workflow operations

  • Structure ingest
  • Relaxation
  • Scoring
  • Ranking
  • Refinement

Material systems

  • Crystals
  • Surfaces
  • Interfaces
  • Defects
  • Grain boundaries
  • Coatings
  • 2D materials
  • Porous materials
  • Amorphous materials
  • Nanostructures

Material classes

  • Metals and alloys
  • Semiconductors
  • Ceramics and glasses
  • Polymers
  • Carbon materials
  • Composites and hybrids

Industries we serve

Energy

battery electrodes · solid electrolytes · fusion first-wall · fission cladding · photovoltaic materials · electrolyzer catalysts

Semiconductors

SiC/GaN · defects · interfaces · dielectrics · interconnects · thermal transport

Automotive

lightweight alloys · advanced steels · hydrogen-compatible metals · magnets · coatings · deformation and fracture

Aero / Space / Defense

thermal barrier coatings · high-temperature alloys · radiation damage · oxidation

Chemicals

catalyst surfaces · membranes · coatings · adsorption and diffusion

Electronics

solder/intermetallics · thermal interface materials · contact materials

Robotics

actuator magnets · structural alloys · joint wear · friction

MedTech / Biomaterials

implant alloys · saline corrosion · coating adhesion · biointerfaces

Metals & Mining

alloy design · phase stability · hydrogen in steel · refractories · corrosion

Deployment

Deployment inside customer-managed infrastructure.

Tensir can run fully inside customer-managed infrastructure using standard containerized deployment, including Kubernetes and Helm. One software interface supports computational work across cloud, HPC, on-premises, and isolated or air-gapped environments.

  • Public / private cloud
  • HPC
  • On-premises
  • Isolated / air-gapped environments

Research data path

Research data can remain within the customer environment. When bursting to cloud compute, execution and data can flow directly between the customer-side Tensir deployment and selected compute infrastructure. No transit through, proxying by, or aggregation on Tensir-controlled servers is required.

Contact

Request a Demo

Tell us about your research pipeline, computational methods, and infrastructure.