Migration Case Study

NeuralSpace enables customers to achieve global reach with linguistic diversity by deploying self-improving custom end-to-end solutions in 100+ languages using groundbreaking AI technology.

The challenge

In their previous environment, the NeuralSpace team had to manually manage and train models across their four products, which was time-consuming, expensive, and really impacting their plans to scale.

One of their key requirements was to reduce this operational overhead, freeing up teams to focus on developing their software features.

So, they began to evaluate migrating to AWS. NeuralSpace realised they would need a trusted partner to help them build out a strategic roadmap and business case and work with their team to validate their thinking. NeuralSpace contacted AWS to discuss and were introduced to Rebura to move the plan forward.

The solution

Our first step was to assess NeuralSpace’s cloud estate. As a migration competency partner, Rebura utilised the SMAP program (now part of the Incremental Workloads program), which provided crucial funding to enable NeuralSpace to move forward.

Through a series of workshops, we were able to understand their current infrastructure, design how it could look on AWS, and provide the running costs for NeuralSpace moving forward. This helped us create a detailed business case for migration, while building NeuralSpace’s confidence in the recommendations and approach.

Our assessment findings included a recommendation that, to deliver on their business objectives, NeuralSpace should utilise Amazon SageMaker.

The findings and detail from the assessment allowed us to move quickly onto the migration itself. We did this using Startup Migration Funding (also now part of the Incremental Workloads program), meaning the majority of the project was subsidised by AWS, removing the cost barrier to migration.

Before completing the full migration, we built out a landing zone for NeuralSpace and created working proof-of-concepts for testing, ensuring Neuralspace experienced zero downtime during switch-over.

Rebura built NeuralSpace’s new infrastructure as code, making it easy for the AWS services to scale as the business requires.

The results

Rebura successfully migrated NeuralSpace’s four products from their old environment onto AWS using SageMaker. This means that their architecture is now:

• Scalable
• Resilient
• Cost-efficient (across implementation and running costs)

Since SageMaker is an AWS managed service, NeuralSpace has been able to reduce their operational overhead significantly. In fact, we expect to see 50% speedup in training time (four days instead of eight).

Today, NeuralSpace is incredibly happy with the outcome.