Oxford Immune Algorithmics leverages science and computational intelligence to advance medical diagnostics and understanding. Their flagship product, Algocyte®, creates personalised health profiles that adapt and learn from individual health behaviours over time.
Challenge
Oxford Immune needed to automate Algocyte’s complex workflow for white blood cell (WBC) image analysis, including model hosting, multi-region deployment, and secure multi-environment management. The solution had to support multiple extraction and classification models, run efficiently at scale, and produce consistent, accurate outputs.
The system required parallel processing capabilities to handle various extraction types, aggregation of results into structured JSON outputs, and a robust CI/CD pipeline for seamless deployment across development, testing, and production environments – all while maintaining high security and operational control.
Solution
Firemind built an AWS-based execution flow and CI/CD pipeline to automate Algocyte’s WBC workflow. Amazon SageMaker Batch hosted and processed models, with AWS Lambda triggering runs as needed. AWS Step Functions enabled concurrent execution of multiple extraction and classification types, while Amazon DynamoDB stored and aggregated results into JSON files in Amazon S3.
The multi-environment CI/CD pipeline streamlined deployment across development, testing, and production, improving security, quality, and collaboration while supporting multi-region scalability.
Services used
- Amazon Bedrock
- AWS Lambda
- Amazon DynamoDB
- Amazon SageMaker
The Results
- 10-day turnaround from project initiation to delivery
- Consistent and accurate WBC analysis results
- Fully automated, multi-environment CI/CD pipeline
- Secure and controlled deployments across regions
