Orchestration layer

Extending your data science tools and infrastrucure

Use the trac platform to:

  • Curate models and data
  • Orchestrate workflows
  • Run models and calculations

Let the platform take care of:

  • Deployment controls
  • Libraries and environments
  • Versioning & lineage
  • Documentation
  • Fine-grained permissions

Take a tour

See how trac integrates your data and compute infrastructure, code repositories and development tools

Universal pattern

Use the trac framework in any environment

  1. 01 - In an IDE

    Use the trac model runtime in your development environment to build portable, production-strength models.

  2. 02 - On your desktop

    A single-user sandbox deployment - ideal for exploration or to replace EUCs and non-production workflows.

  3. 03 - In the cloud

    For AWS, Azure and GCP we offer a simple marketplace solution or a Kubernetes pattern, for production workloads.

  4. 04 - On-premises

    Various on-premises deployment patterns can also be supported depending on your infrastructure availability and needs.

Download the free desktop version and start exploring today.

Try Community Desktop

Unique design

Automating controls from the ground up

Self-describing

Built around a semantic data model that catalogues and describes every asset, process, and calculation.

Self-documenting

Time-consistent action and update recording - a complete version history of the whole platform.

Stateful

Every version of every asset and every prior state of the platform is available to each user, all the time.

Explore control framework

Built for self-service

Democratising model use

The unique control environment empowers model owners and model users to self-serve, free of change risk — manual controls and traditional access restrictions cease to be the bottleneck.

Explore user personas
Description of image
Descriptive alt text

Our philosophy

Built on open-source foundations

We publish trac's core analytic services as an open-source project (TRAC D.A.P.) via FINOS. The trac model runtime is also available to download from pypi.org.

Explore open-source

The trac ecosystem

Flexible integration options

Integrating your data and compute infrastructure, code repositories and development tools to create a unified ecosytem in which to build, deploy and use models.

An append only data model ensures job repeatability, so trac must control write access to primary storage. For cloud deployments this would be regular bucket storage (S3, GCS or Azure Blob).

We provide a Kubernetes pattern with single-node and distributed (Spark) capabilities for cloud. Integrations with other compute services are available for on-premises.

Models are stored in an external repository and the code is accessed dynamically at runtime. Git and Nexus are supported natively, custom repositories are accommodated with plug-ins.

Deploy the trac runtime (python) to your IDE and build models that can be deployed to production with no code modification.

For batch data import and export, connectivity is available for most common technologies including; object storage, file-like locations, databases, and SQL-like locations.

Plug into your SSO mechanism with roles assigned in the AD and fine-grain permissions configured in trac. Azure AD, SAML and OpenID are supported natively, other solutions are accommodated with plug-ins.