Back to our work
Platform engineering

A new foundation for an analytics platform

Cloud-independent infrastructure, event-driven services and a faster, cheaper data layer for an established learning analytics platform.

Period
Aug 2024 - May 2026
Our scope
Platform transformation
Engagement
Direct client engagement
Platform architecture before and after modernisation Cloud vendor lock-in to Cloud independent. Monolith to Microservices. Manual provisioning to Infrastructure as code. Multiple sources of truth to Single source of truth. Synchronous calls to Asynchronous messaging. Standard reports to LLM-assisted analysis. The added analysis applies to selected reports. Before After Cloud vendor lock-in Cloud independent Monolith Microservices Manual provisioning Infrastructure as code Multiple sources of truth Single source of truth Synchronous calls Asynchronous messaging Standard reports LLM-assisted analysis Platform architecture before and after modernisation Cloud vendor lock-in to Cloud independent. Monolith to Microservices. Manual provisioning to Infrastructure as code. Multiple sources of truth to Single source of truth. Synchronous calls to Asynchronous messaging. Standard reports to LLM-assisted analysis. The added analysis applies to selected reports. Before After Cloud vendor lock-in Cloud independent Monolith Microservices Manual provisioning Infrastructure as code Multiple sources of truth Single source of truth Synchronous calls Asynchronous messaging Standard reports LLM-assisted analysis
Explanatory diagram illustrating the engineering approach.

Key results

  • The cost of the analytical data layer roughly halved after moving workloads from Amazon Redshift to PostgreSQL clusters tuned for OLAP queries.
  • Average response times for analytical queries decreased, giving users a noticeably better experience.
  • Changes to the administrative database typically reached replicas in other regions in 30-300 ms. Customer data remained in its designated region.
  • A cloud-independent stack designed for deployment with any cloud provider or on the client’s own servers, with infrastructure described in Terraform.
  • Pull-based GitOps delivery with a controlled rollout of every deployment across regions.

The challenge

IntelliBoard serves customers in several regions of the world, and the law does not allow customer data to be moved between regions. Its infrastructure was tied to AWS and partly managed through static configuration edited by hand. The product was a PHP monolith, analytical queries ran on Amazon Redshift, and builds and releases were pushed from Jenkins, with the code hosted on Bitbucket.

Our contribution

  • Moved the platform from AWS to Oracle Cloud Infrastructure and described the infrastructure as code with Terraform.
  • Standardised on components that also run on premises: Keycloak, GitLab CE, Harbor, PostgreSQL with pgvector, MinIO, Apache Kafka and Redis.
  • Split the PHP monolith into services connected through Kafka domain events. Part of the services moved to Go, and new machine learning (ML) services were developed in Python.
  • Replaced Amazon Redshift with fault-tolerant PostgreSQL clusters tuned for OLAP queries: a primary with replicas and automatic failover.
  • Integrated large language models (LLMs) to provide supplementary analysis in selected analytics reports.
  • Developed a specialised, fault-tolerant and secure change data capture (CDC) solution for the administrative database. It reads changes from the write-ahead log (WAL) of the primary database and delivers them to replicas in other regions over HTTPS in near real time. Customer data is never replicated outside its region.
  • Replaced push-based Bitbucket and Jenkins pipelines with GitLab CE and Argo CD. GitLab runners build and test; Argo CD deploys from declarative descriptions in dedicated repositories using Helm charts developed for the company.

What the work delivered

Over almost two years of direct work, the platform gained a cloud-independent foundation, a service architecture it can keep evolving, a cheaper and faster analytical data layer and a controlled multi-region release process.

More engineering stories

View all work
Computer vision

Computer vision platform

A computer vision platform that takes a project from data storage and annotation through training to live inference. Our team built two applications to demonstrate the full workflow.

Applied ML

Predictive learning analytics

A service that trains a model on each institution’s data, explains its predictions and analyses text, topics and sentiment. It launched in 2021 and was extended in 2025-2026.

Data engineering

Enterprise data warehouse

A BigQuery warehouse that brought sales, e-commerce, web analytics and sales-force data together, with Power BI reporting and market basket analysis for the US and Canada.

Let’s build something useful

Bring us the
hard part.

A new AI capability, a complex data problem or a platform that needs to evolve. Tell us what you’re working on.

Discuss your project

Start a conversation

info@ai-labs.tech

aiLabs Innovations FZ-LLC
Compass Building, Al Shohada Road, Al Hamra Industrial Zone-FZ, Ras Al Khaimah
United Arab Emirates
Licence No. 47014246

For an initial conversation, share your goal, the current system and the challenge you want to solve.