The client engagement process begins with an on-site executive briefing, which provides an overview of Camgian’s Cognitive Computing Services, and the identification of a high-priority client case study.  Next, technical diligence is conducted by Camgian personnel to evaluate the feasibility of developing a robust machine learning model based on sample data provided by the client.  Results of the diligence effort are provided to the client and if determined feasible, a proposed plan to develop and demonstrate an integrated data platform (IDP) and model prototype(s) are included as deliverables.
 

Typical prototyping programs begin with building an IDP to support machine learning model training and testing.  This includes identifying the requisite data sources for modeling activities and performing data engineering to clean, structure, format, normalize and integrate the data sources as required.  Concurrently, based on the quality of the data and problem set, Camgian data scientists identify the best machine learning approaches to meet client objectives and desired capabilities.  Prototype models are then trained, tested, and optimized using Camgian’s application pipelines and tools and demonstrated to the client, along with the IDP, as deliverables for the prototyping phase of the program.      
 

Camgian leverages an agile development process which includes sprints comprising short design and development cycles. Stand-up meetings with client personnel are conducted at frequent intervals to review both the results of the previous sprint and acquire feedback on the design of the next sprint cycle. This agile development approach and close client engagement ensure that operational needs and objectives are continuously reviewed and integrated into the overall development program.

The production phase of an engagement includes working closely with client operations and IT personnel to stand up a real-time implementation of the IDP and model prototype(s).  This includes integrating with the requisite client IT infrastructure and operationally deploying a prioritized set of models.  Once established, the production IDP provides the foundation for integrating new data sources and deploying large numbers of future algorithms and models at scale.  This flexibility provides clients the capability to continuously capture additional economic benefits with the deployment of new models operating on expanded versions of the same core data infrastructure.