Expanding Outreach Platform with Microsoft Azure AI Systems for Aviation Manufacturer

Background and Challenge Story

Finding A Way to Implement a Consolidated Platform to Build Future AI Projects

The client is a leading manufacturer of business aircraft in Canada. They prioritize innovation and are committed to providing contemporary aviation products powered by the latest technologies.

The client’s primary objective with this partnership was to establish a consolidated platform on Azure to develop AI-related models and projects, while showcasing the success of past projects internally. This process involved proving the platform’s validity by migrating and highlighting an existing forecasting solution, as well as demonstrating the functionality of MLOps (Machine Learning operations) and automated promotion pipelines integrated into the platform to manage model accuracy and re-training triggers. The client plans to build every AI-related project on this platform moving forward.

Solution Story

Four-Phase Project to Establish a Consolidated Platform within Azure

For about a year prior to the project, the client was in conversation with Microsoft that involved proposal discussions. The project was broken into four phases:

Phase I: Use Case Summarization and Prioritization

The client engaged in extensive conversations and interviews with Adastra, spanning various divisions such as HR, manufacturing, and IT. These discussions aimed to help Adastra comprehend and visualize the client’s highest value potential use cases across all divisions. Following these discussions, the client received a detailed matrix outlining each identified use case.

Over a two-week period, the client was provided with recommendations for project implementation, including suggested tools, potential concerns, and general notes for all scenarios.

To prioritize the identified use cases, Adastra and the client focused on two key factors: the potential business value impact and the feasibility of implementation. Based on these criteria, each use case was assigned scores for impact and feasibility.

Phase II: Initial Design and Architecture Build

During this phase, the client received Insight Zones – AI platforms built in the Microsoft Azure cloud. Each Insight Zone served as a dedicated hub for a specific business unit, laying the foundation for all future AI projects.

The client’s vision for the Insight Zones was clear: they needed to seamlessly implement the highest priority use cases while remaining adaptable for future needs and customizations. Moreover, the Insight Zones had to be fully automatable with a streamlined deployment process, facilitated by the comprehensive developed of CI/CD pipelines.

The architectural blueprint for developing the Insight Zones comprised three environments – a dev (development) environment, a QA (test) environment, and a production environment – each connected to various data sources. The entire platform was built inside Azure and involved comprehensive knowledge transfer.

This advanced analytics infrastructure provided the client with many benefits, including:

• Reliability and consistency across all environments.
• Invaluable insights into company data through Insight Zones.
• Reduced costs and accelerated deployment times.
• Seamless infrastructure deployment through automation.

By the end of this phase, the entire platform had been developed. The Insight Zones provide the client with a fully automated infrastructure deployment process applicable across every business unit, from development to production. Before transitioning from this phase, Adastra ensured that the client was comfortable deploying these new business units.

Phase III: Implementation of Existing AI Solution on Newly Built AI Platform

During this phase, the client’s pre-existing use cases, developed by their internal data science team, were transitioned to the newly established AI platform. Leveraging the newly built CI/CD pipelines, the client facilitated ongoing development and deployment of the solution.

Post-implementation, the client is now poised to stabilize the solution and conduct performance evaluations against alternative methodological approaches to validate the efficacy of the proposed model paradigm. A significant amount of refactoring of the existing solution was also completed during this phase, encompassing enhancements in runtime speeds for heightened efficiency and the integration of data science best practices.

Phase IV: Consolidation of Documentation Across the Entire Project Journey

The final phase of the project involved consolidating documentation related to both the newly established AI platform and any modifications made to existing AI infrastructure during Phase III. Key deliverables provided to the client encompassed comprehensive user guides and runbooks, detailing the steps for platform onboarding and user orientation.

Furthermore, a series of user guides were provided in the form of video tutorials, outlining the utilization of the new platform, comprehension of the design and deployment processes, and facilitation of machine-learning projects.

Following the completion of all four phases, the client expressed the following requirements:

• Access to output documents from Phase I, delineating the definition, input, and prioritization of each identified use-case.
• Visibility into the process involved in working with their pre-existing solution during Phase III.
• Provision of video and document walkthroughs/guides for platform usage during Phase IV.
• Inclusion of any additional documentation outlined in the project plan as part of Phase IV.

Benefits Story

Lower Costs, Saved Time, and Improved Reliability & Consistency

Adastra’s solution has positioned the client to integrate Azure AI into their AI-building platform, streamlining operations and ensuring the success of both past and future projects.

The collaboration between Adastra, Microsoft and the client resulted in the following results:

• Time and cost savings.
• Accelerated deployment of advanced analytics infrastructure.
• Reliability and consistency through “Insight Zones.”
• Fully automated development processes.
• Extensive insight into the value of various use cases.
• Anticipated future time savings through streamlined implementation processes.
• Flexible platform adaptable to future use case scenarios and customizations.

Moving forward, the client aims to continue benefiting from the time and cost advantages provided by their new automated and expanded outreach platform. They also plan to migrate previously completed projects to the new outreach platform to validate their success. Adastra and Microsoft are ready to support the client’s team in all future endeavors.