For the first project, I was tasked with reviewing the existing Nibble setup and:

1. provide recommendations for best practices regarding their infrastructure and security

2. ensure their API could handle high-volume traffic effectively

After conducting an analysis, I recommended migrating their Lambda API to the Serverless Framework and adopting Terraform for managing their AWS infrastructure. I implemented the configuration along with scripts to facilitate easy deployment on a per-environment basis. Additionally, as part of the migration, I transitioned their production database to Aurora Serverless to enhance scalability.

For volume testing, I set up a JMeter script to simulate expected client patterns, including drop-off scenarios and various client flows. During load testing, I identified limitations with AWS Lex’s transactions per second (TPS) limit and collaborated with AWS to increase the TPS limit for the Nibble account.

For the second project, I was tasked with incorporating Rasa into the Nibble platform to replace Lex. For cost efficiencies, we also decided to migrate from Fargate to EKS.

The DevOps transformation programme delivered a more reliable, scalable, and maintainable platform for Nibble. The migration to the Serverless Framework and Terraform standardised infrastructure and deployments across environments, enabling predictable, repeatable releases. Aurora Serverless and the performance-testing programme ensured the API could scale to meet peak client traffic while maintaining stability.

The JMeter-based load-testing framework provided ongoing assurance that the system could handle both expected and stress-level volumes. Performance bottlenecks were identified and addressed ahead of production rollout. During this process, transaction-per-second constraints within AWS Lex were highlighted and resolved in collaboration with AWS.

The second phase of work successfully introduced Rasa and migrated the workload to EKS, positioning the platform for future product development and cost-effective scaling. Following these improvements, the company expanded its capability by appointing a full-time DevOps engineer. I was invited to take the role permanently, but I chose to remain in contracting.

Overall, the platform evolved from a manually deployed early-stage product into a robust, cloud-native service capable of supporting live production growth.