Technology Platform - Izengard Unleashed
Core Platform
The core technology platform of Izengard’s horizontal services layer is built on modern scalable architecture including the following:
Front End
- Java with HTML 5 and CSS 3
- Java/Python Rest API calls to middle integration layer
- Low code interface
Database Layers
- Modern Data Fabric / data mesh, supporting federated, tokenizing it for common functional processing and supporting SQL, NoSQL processing for structured and unstructured data. Full data catalog and data lineage supported as well as full distribution of logic and use of in-memory storage.
- A Personally Identifiable Information Vault (which stores masked data) but is embedded on-site at your own premise or on the cloud data fabric / data mesh.
- Scalable Casual AI ML-Ops platform working on top of the data fabric with strong pipelines, model management and version control and an array of modelling techniques. Izengard supports Causal AI algorithms for ensemble modelling, structured, unstructured modelling and re-inforcement learning.
Back End Serverless Processing
- Hi-speed messaging layer able to scale to 5000tps
- Containerisation of all components
- Multi-user process handling and deployment of robots
- Scalable compute engine
- Multi-tenancy with 99.999% RTO.
- Optimized on-premise environments where required and initially set up for load balancing by doing careful capacity planning and performance considerations. Partnerships with several bare metal providers which can result in optimized hardware, software and network capabilities that enhance the on-premise environment
Sandbox Environments
- For DEV, SIT and UAT environments.
- For Model Building environments
- Synthetic Data Generation
- For Line 3 internal audit testing
- Model Validation
- Controls Testing
- Risk Rating
- Control Libraries
- Validation Reporting
- Statistical Modeling
- Machine Learning Modeling
- Rules Based Modeling
- Replicated as smaller versions of above platform components and needed horizontal components.
Horizontal components
Izengard has several key horizontal components that allow us to integrate the 3 domains. Amongst the most important in the architecture are:
- A data fabric / data mesh technology allowing our data to reside where it exists, but bring it together for via a knowledge graph metadata approach for common processing as well as a PII vault with masked data.
- Izengard Mind Map = advanced entity resolution layer, with contextually adaptive profiling combining resolution to avoid ID fraud and context to combine risk indicators
- An intelligent log ingestion layer to handle both customer and employee digital footprints, so that these can be analyzed for unusual behavior.
- A case management and investigations layer that combines data from the 3 domains and presents a holistic view of risk. This layer also provides a lot of automation as volume of transactions or logins rise.
- A risk aggregation and scoring layer which allows our contextually adaptive engine to pinpoint, prevent and detect highest risk events.
- An advanced anomaly detection layer that works with the advanced entity resolution and risk aggregation/scoring layer to detect high probability events and decrease the chances of false positives by combining context with the risk events.
- An advanced BPMN compliant workflow orchestration engine which integrates and controls all activities and links the 3 domains with the case management and investigations layer.
- A machine learning ops platform that co-works with our data mesh / data fabric but provides for Causal AI algorithms and approaches
- Izengard Data Detective, reviewing device level behavior from individuals, corporates etc and examining anomalies to deter ID fraud, ID theft, social engineering, phishing attacks etc.
- Intelligent robots – applying forensic data preservation techniques and carrying out low level linkages between pieces of evidence in the background and co-working the Graph database/network analytics components within Izengard. Robots are additionally implemented to carry out data reconciliation as well as case dispositioning tasks at a lower level.