Deep Learning A.I. Data Fabric Architecture

Schedule Demo

A Data Architecture for Deep Learning A.I. Analytics

At Vyasa, we realized that novel deep learning A.I. analytics approaches require access to many different kinds of data and the ability to generate specific trained models for each type. As a result we developed Layar as a “data fabric” approach to novel deep learning A.I. data analytics.

Layar’s advanced BERT-based A.I. analytics text modeling engine can be trained to derive meaning directly from document repositories which enables organizations to power fundamentally new use cases that do not require linked data, ontological modeling or regex word matching.

Layar is a secure and scalable solution that can be deployed on cloud or on-prem and offers a full suite of analytics modules, data management pipelines, and APIs designed to provide your company with an elegant and sophisticated architecture to glean insights from your data.

Tailored Data Fabrics

Manage and connect data across diverse environments by weaving it together into a client-specific data fabric that can tap into a mosaic of private and public pools of data without a dependency on standardized controls or preprocessing.


Hone in to specific sections of unstructured and structured content (such as the executive summary of a PDF, or the metadata for a JPG) to drill down on dataset parameters and model refinement.

AWSGoogle Cloud PlatformMicrosoft Azure
On-prem StackLayar API

Flexible Deployment Options

Layar can be deployed as a fully containerized Helm/Kubernetes stack on cloud environments including Google Cloud, AWS and Azure or on-premise.

API / Microservices Architecture

Layar provides an extensive set of API calls and microservices to enable flexible deployment and custom or third-party application integration.

Breast Cancer Detection

Detection of breast cancer on screening mammography is challenging as an image classification task because cancerous tissue only represents a small portion of the tissue in the image. Cortex rises to the challenge with localized tiling to deliver state of the art results.

Crystal Morphology Classification

Microscopic images of drug crystals are generally evaluated and classified by subject matter experts, a bottleneck in the process that Cortex can help solve.

Identifying Emergent Companies and Patents

Automated identification of emergent technologies, patents, and companies from unstructured text can be challenging. Synapse can uncover even the most obscured similarities between technologies hidden deep in documents, making discovery efficient and effective.

Data Ingest for Public & Private Datasets

Layar leverages containerization and clustering technology to ensure seamless and efficient scaling and integration of an unlimited number of data repositories. Layar’s data fabric can process a variety of disparate data types, including but not limited to: XML, PDF, Sharepoint, Twitter, RSS, XLS, TXT, CSV, TSV, DOC, DOCX, and so many more.

We also offer access to the Layar Data Catalog, our off-the-shelf collection of Layar data sources that we use to continuously fine-tune our deep learning models.

The National Institute for Health and Care Excellence (NICE)PubMed AbstractsPubMed Central (PMC) Open AccessClinicalTrialsUS Patent Office (USPTO)

Dynamic Compute Technology

Deep learning models typically require substantial computing power for pre-training and ingestion of novel texts, which can be expensive if a company attempts to build and train algorithms from scratch. Layar avoids this with its hybrid GPU/CPU architecture and GPU Smart Switching capabilities, which allow deep learning training and model utilization to run seamlessly and efficiently.


Vyasa Analytics has over a decade of data analytics expertise available to design and deploy deep learning software. We offer support from our experienced engineers and solution architects, who can advise you on strategy, implementation, and development of our software to optimize its functionality for your use cases.

Contact us

Get In Touch

You can email us at or use the form below, your choice.