The client set out to integrate a third-party scientific platform with their systems by 2023. Five months into the project, Janea Systems was brought on board to lead the client’s in-house engineers and external contractors’ team.

The project involved migrating the source code of a complex app consisting of nine microservices, which was originally hosted in a third-party AWS environment. Our task was to adapt it to a new infrastructure managed by the client, where network configurations and infrastructure components were strictly controlled. This required modifying authentication systems, migrating data, and maintaining the integrity of a complex architecture.

Infrastructure Optimization

As our engineers examined the infrastructure, they discovered that it ran on self-managed Kubernetes clusters, adding unnecessary overhead and complexity. The deployment wasn’t optimized for the client’s scale and could've benefited from transitioning to AWS-managed Kubernetes (EKS). We ensured proper repository management by storing Docker images in the private repository (JFrog Artifactory) while navigating multiple approval processes.

Compliance & Security Alignment

Adherence to regulatory requirements was critical. The application must comply with CFR 21 Part 11 and GXP regulations, including Good Manufacturing Practices (GMP) and Good Laboratory Practices (GLP). Our team provided extensive documentation, passed Target State Architecture assessments, and followed quality assurance processes within the internal SDLC.

Authentication System Overhaul

The authentication system overhaul involved replacing Okta SSO with the Ping Federate authentication system. Our engineers fully rewrote the user access management logic to align with the client’s authentication framework.

Technical Debt Management

The inherited codebase was a mix of JavaScript (TypeScript, CoffeeScript), Python, and Terraform, with outdated components such as AWS Lambda functions running on Node.js 12. Upgrading and refactoring these elements was necessary to ensure stability and maintainability.

Signals Notebook Improvement

Our team identified a functionality gap in the Signals Notebook – an electronic lab notebook (ELN) solution by Revvity. Through our interactions with scientists, it became clear that users required greater workflow automation and the ability to leverage Python beyond Jupyter Notebooks. Scientists relied on this tool for data processing, yet it couldn’t handle datasets larger than 2,000 rows. This limitation forced them to rely on manual workflows using Excel.

Our engineers developed the Extensible Data Connector from the ground up to address these limitations. The solution lets scientists create custom scripts for data processing. Given the widespread preference for Python-based solutions, user feedback has been overwhelmingly positive. On top of that, users requested our team to add more workflows and frameworks, including Dash and Streamlit.

Additionally, the SQUIRREL platform was identified as a candidate for integration with the Extensible Data Connector. SQUIRREL is a Python tool for small data processing scripts that can be scheduled as periodic jobs, a functionality we recently added for further workflow automation.