A major European financial institution needed to modernize from legacy 1980s statistical software to Python-based risk modeling while operating under stringent security, data residency, and regulatory compliance requirements.
The Anaconda Platform provided centralized Python development with enterprise governance, curated package repositories, automated vulnerability detection, and complete audit trails across global data centers.
Global Financial Institution Secures Risk Modeling Transformation While Scaling to 300 Python Developers
A major European Financial Institution was losing business to competitors. Small and medium enterprises needed lending decisions faster, and the financial institution’s decades-old proprietary statistical software couldn’t keep pace. As customers took their business elsewhere, the pressure mounted: modernize risk analytics capabilities to leverage Python and machine learning, but do it while operating under some of the most stringent security and regulatory compliance requirements in financial services.
The stakes couldn’t be higher. Risk models aren’t optional analytics projects at this financial institution. They’re the capital models, lending decisioning systems, and stress testing frameworks that regulators require for the institution to maintain its license to operate. Some models take 18-24 months just to clear regulatory approval. If their systems were compromised for more than three days, the financial institution could lose its license to operate, disrupting payments for hundreds of thousands of people and critical funding for major corporations.
The Challenge: Modernization Meets Compliance
When the financial institution’s technology team began supporting its Analytics division over 10 years ago, modelers relied almost entirely on proprietary statistical software from the 1980s. These modelers built models with minimal IT involvement and virtually no software engineering practices.
As business pressure mounted to deliver lending decisions faster, particularly for small enterprises taking business to competitors, the limitations became critical. The marketplace was moving toward Python, R, and machine learning techniques the legacy platform couldn’t support.
But the financial institution couldn’t simply install Python and let modelers start coding. Strict data residency requirements meant customer data couldn’t leave its approved jurisdiction.
This meant desktop development was impossible. Models needed to be built on centralized infrastructure where data could remain in its approved location, with complex access control groups enforcing who could access what data across different teams, jurisdictions, and model types. Model developers, calibration specialists, and review boards (in different locations) required carefully managed access to the same data without the ability to download it locally.
The security requirements were equally stringent, as the financial institution maintains a zero-tolerance policy for high risk vulnerabilities. To comply with this standard, the technology team spent approximately 60% of their time identifying and remediating security issues before deployment. Allowing analysts to install arbitrary packages from open repositories risked introducing malicious code into systems processing customer financial data.
The financial institution needed a platform that could enforce data boundaries, manage complex permissions across global teams, provide curated packages from trusted sources, and maintain complete audit trails for regulatory compliance while enabling modelers to leverage modern Python capabilities.
Finding the Right Foundation
The financial institution didn’t have time for a lengthy platform evaluation. Business pressure was mounting, and they needed a solution that could be deployed quickly without compromising their stringent security and compliance requirements.
The answer was already running in their environment. Pockets of the organization were using Anaconda, and another division had successfully implemented Anaconda Enterprise in a server environment. Rather than starting from scratch with an unknown platform, the technology team could leverage an established solution that had already cleared security reviews.
The Anaconda Platform offered exactly what the financial institution needed: enterprise-grade governance and security capabilities, and immediate productivity for modeling teams.
The Solution: Enterprise Python Without Sacrificing Security
The Anaconda Platform addressed each of the financial institution’s critical constraints:
Solving the Data Residency Challenge: The platform’s centralized architecture enabled the financial institution to deploy instances in their own data centers across the globe. Data could remain in its approved jurisdiction while modelers accessed it through the platform, eliminating the desktop development problem that would have required moving sensitive customer data across borders.
Managing Complex Permissions at Scale: With Active Directory integration, the platform could enforce the financial institution’s complex access control groups without requiring modelers to understand the underlying complexity. Model developers, calibration specialists, and review boards in different locations could access exactly the data they needed, nothing more, with complete audit trails of who accessed what and when.
Reducing Security Vulnerabilities: Rather than allowing unrestricted access to open repositories where analysts might inadvertently download malicious packages, Anaconda’s curated distribution provided vetted, tested packages that satisfied IT security requirements. “If we were to just use PyPI repo, the problem is that it puts a lot of onus on managing dependencies on yourself, and you’ve got to have a lot of engineering around that,” the technology leader explains. The platform’s automated vulnerability detection and provenance tracking gave the security team visibility into the software supply chain, helping address the vulnerabilities that consumed 60% of their time.
Enabling Faster Model Development: Isolated Conda environments allowed modelers to work on multiple models simultaneously without dependency conflicts, each with its own package versions. The platform’s project construct enabled teams to quickly spin up collaborative workspaces, add members with appropriate access, and move from analysis to production-ready code—all within a single governed environment. This eliminated the time-consuming handoffs between statisticians and IT teams that had slowed lending decisions and cost them business.
Scaling from Adoption to Enterprise Standard
The initial rollout succeeded dramatically. Within 18 months, modeling teams had created approximately 500 projects across all server environments. While this demonstrated strong adoption, it also created governance challenges.
“We were a victim of our own success,” the technology leader recalls. Many projects were abandoned training exercises or temporary explorations. As the platform evolved, the team implemented more rigorous governance with structured naming conventions tied to the formal model registry.
Platform upgrades also delivered critical technical improvements through Kubernetes architecture, which provides dedicated capacity per session and eliminates resource contention issues.
Supporting Mission-Critical Model Development
Today, 300 active modelers use the Anaconda Platform across the financial institution. The platform underpins a sophisticated workflow satisfying both regulatory requirements and operational needs.
For capital models requiring independent approval, modelers must document their complete analytical journey: data sourcing, quality assessment, factor analysis, methodology selection, and validation. With the legacy software, this meant producing separate documentation in Word and PowerPoint, disconnected from the actual analysis. Jupyter notebooks helped this process by combining executable code, narrative explanations, and visualizations in a single reproducible document that independent review teams and regulators can replay and verify.
The addition of VS Code and integrated development environments enabled the critical next step: moving from analysis to production-ready code. Model engineering teams now work directly with modelers within the same platform, embedding software engineering practices into model development. “What you have to stop at is the fact that you deploy into production,” the technology leader emphasizes. “You have to make sure that whatever you’re producing, you can run it operationally.”
This end-to-end workflow within a single platform replaced the previous handoff process where statisticians would analyze data in the legacy software, document their findings separately, then pass everything to IT teams to rebuild for production, a process that introduced errors and delays.
Once models pass independent approval, they’re packaged into Docker containers with defined library dependencies from their conda environments. The platform provides end-to-end traceability, linking model IDs to their code repositories, container artifacts, and production deployments with complete audit trails.
Managing Risk in an Open Source World
Organizations across industries have embraced open source Python for its innovation and flexibility. But for regulated financial institutions, the challenge is different: How do you leverage the power of the open source ecosystem while meeting stringent security and compliance requirements that prohibit unrestricted access to public repositories?
The Anaconda Platform’s security capabilities proved essential for threading this needle. Beyond providing curated packages with signature verification, the platform enables the comprehensive vulnerability management the financial institution requires to maintain its zero-tolerance policy for critical vulnerabilities in production.
Every week, vulnerability scanning tools identify new security issues across the technology stack, requiring constant remediation work. “We get more and more vulnerabilities, and we have to constantly address them all the time,” the technology leader notes.
The platform’s provenance tracking and automated vulnerability detection provide crucial visibility into the software supply chain, enabling the financial institution to rapidly assess and respond when issues emerge in the Python ecosystem while maintaining the audit trails regulators require.
A Cultural Transformation: From Analysts to Engineers
Perhaps the most significant impact extends beyond technology to organizational culture and business results. The financial institution is fundamentally changing how models are built, moving from pure statistical analysis to end-to-end engineering ownership.
The business impact has been substantial. The platform’s integration with the data platform enables teams to start work immediately. “As Anaconda is hooked into the data platform, then we can simply allocate a new project and away we go,” the technology leader explains. “It allows the business to get off the ground quick and collaborate.” This speed matters because the financial institution had been losing business to competitors, where small organizations needed quick loan decisions and were taking their business to other banks instead.
Historically, modelers would analyze data in the legacy software, develop predictions, document findings in separate Word and PowerPoint files, then hand everything off to IT teams for operationalization. This created costly disconnects. Models might rely on data sources unavailable in operational systems or use techniques impossible to run at scale. IT teams spent significant time understanding the analysis and rebuilding it for production, introducing both errors and delays.
Now, modelers must think about operational deployment from the start. The Anaconda Platform enables this transformation by providing tools data scientists need (notebooks for exploration, visualization libraries, statistical packages) while supporting engineering practices the organization requires through IDEs, version control integration, and reproducible environments. The shift is dramatic. Previously, modelers produced code that was difficult to understand, impossible to componentize, and couldn’t be deployed to production. Now they build modular, containerized models with engineering discipline embedded from the start.
Looking Forward
The financial institution continues evolving its analytics capabilities, with plans to extend into Google Cloud Platform’s Kubernetes environment. The challenge remains: balancing innovation velocity with complex data residency requirements across its global footprint.
The Anaconda Platform delivered exactly what this institution needed: the confidence to innovate with open source at enterprise scale, backed by the security, governance, and reproducibility of their business demands. From supporting a small initial team to enabling 300+ modelers across global operations, the platform has become the foundation for the financial institution’s risk modeling transformation.
Ready to scale your AI development with enterprise-grade security and governance? Discover how the Anaconda Platform can transform your organization’s approach to AI with open source. Contact our team for a personalized demonstration.
Anaconda Platform
The Only Unified AI Platform for Open Source