Data Science

What Our Capabilities Are

DATA SCIENCE

SOFA

SOFA: Simulation of Federated Applications​

Call applications like functions:

  • Legacy tools, simulations, scripts, etc.​
  • Tools containerized with dependencies​.
  • Simple JSON/HTTP client interfaces​.

Compose meta-applications:

  • Polyglot, multi-operating system​.
  • Fully asynchronous parallel execution​.
  • Tools may execute more tools​.

Focused on supporting analysis​:

  • Easily create custom drivers, post-processing​.
  • Execute and query data from other tools​.
  • Some UQ, design, optimization methods​

The Simulation of Federated Applications (SOFA) software is a platform which provides “containerized functions as a service” via a simplified REST API whereby containerized tools may be registered with SOFA and called arbitrarily via simple POST/GET commands. SOFA was designed for the Air Force Research Laboratory (AFRL) to support scientific analyses utilizing legacy simulation tools which are known to be difficult to install, operate, and/or parallelize, and for easy incorporation of ad-hoc tools such as post-processing scripts and ML training. ​

Contact Us

SOFA 2.0 has been designed from the ground up to allow tool and client integrations to be implemented with as little code as possible. Additional tooling has been created on top of the SOFA platform such as SOFA-Workflow and SOFA-Builder. SOFA-Workflow is a tool suite created to provide a visual interface for assembling and executing complex analyses. SOFA-Builder provides a GUI for creating and registering new containerized tools and analysis drivers with SOFA.​

SOFA has drivers for analyses such as Uncertainty Quantification, Optimization, and Design of Experiments out of the box, and allows for additional analysis drivers to be implemented via the same simple interface as other containerized tools. SOFA is backed by S3 (or Minio) artifact storage and a NoSQL database to store inputs and results data. ​

To date, SOFA has been utilized for numerous concept development analyses using tools such as AFSIM, AJEM, CWS, EPIC, IWEA/Endgame Framework, and Missile DATCOM, among others.​

WHAT IS DEVSECOPS?

DevOps is a software practice that aims at unifying software development (Dev) and software operations (Ops). The main characteristic of DevOps is to advocate automation and monitoring during all steps of the software development lifecycle (SDLC). It aims at shorter development cycles, more stable deployments, and increased testing. DevOps, unfortunately, is not enough and will not catch every issue for it is missing the security component, and that is where DevSecOps comes into the equation.

DevSecOps is a methodology used to automate the integration of security at every phase of the SDLC. The purpose and intent of this methodology are to build on the mindset that everyone is responsible for security. Typically, in the SDLC, security is tacked onto the end, and the responsibility is put on Quality Assurance (QA) to find any issues. Now this workflow comes with some notable drawbacks in terms of both time and resources. One method used to eliminate this bottleneck is the inclusion of security, which can be beneficial to the team. This methodology allows for rapid software delivery and produces code that is of production quality.

Software engineering is constantly evolving, and Anyar is partnering with the AFRL, other agencies, and corporate partners to develop CI/CD pipelines and Software Factories to improve the quality of the products our government customers develop using tailored DevSecOp software technology stacks.

WHAT IS CLOUD COMPUTING?

Cloud computing is the practice of using a network of remote servers hosted on the internet to store, manage, and process data rather than a local server or personal computer. Cloud computing provides various benefits, such as utilizing the power of cloud resources and being able to deploy software on a global scale.

We believe that cloud computing is the future. To keep up with an ever-increasing interest in utilizing legacy applications in cloud computing environments, Anyar has partnered with AFRL and AFLCMC to migrate existing legacy multi-OS applications and workflows into SOFA’s container based workflow system using several DoD cloud and networked environments.

JMP-SEAT

Anyar’s JMP-SOFA External Analytics Tool provides a streamlined user experience for the JMP analyst to utilize features of SOFA. Written in JMP’s scripting language, JSL, this tool leverages JMP’s capabilities for the Design of Experiments to launch SOFA processes. With simulation results automatically aggregated and populated into data tables, analytics can ensue using typical JMP methodology.

MACHINE LEARNING

We strive to integrate modern AI/ML methodologies for our DoD customers in all areas of operations research. Anyar has multiple ongoing efforts centered around applying advanced machine learning techniques for MS&A, including high-fidelity penetration model surrogation using physics-informed neural networks, self-optimizing 6-DOF models using reinforcement learning, and trade space evaluation via iterative uncertainty quantification and regression modeling.
Contact Us