Architecture Guiding Principles

We begin by aligning our deliverables to the business objectives defined by the REDs. We deliver iteratively, incorporating continuous customer feedback, and validate our outcomes against the business’s success criteria, scaling only what creates proven value.

Impact-driven, evidence-based delivery ensures we focus our time and investment on what truly matters. Our patients are waiting.

We prioritize reuse by evaluating existing solutions, starting with open-source products, then commercial off-the-shelf solutions, and only considering building from scratch as a last resort, while articulating the architectural rationale when reuse is not viable. Evolving towards modular, loosely coupled, and interoperable components that can be reused and composed ensures accelerated delivery across domains.

Reusability mindset reduces duplication and improves ROI, while modular design enables agility, interoperability, and long-term scalability.

We strive for applying cloud where it makes sense, i.e. cloud-first, cloud-native, SaaS-first, AI-enabled, API-driven architecture that enables rapid innovation and scalability. * Cloud-first means we prioritize the cloud, but are not cloud-only. * Cloud-native means we leverage cloud-providers specific capabilities as an accelerator, rather than being cloud agnostic. * AI-ready means we increasingly embed AI into how we design, build, operate and govern our systems, while also ensuring our systems are AI-ready.

Enable rapid innovation and integrated ecosystems, and create a seamless environment where machine learning and artificial intelligence technologies can thrive.

We automate repeatable, manual, labor intensive work across infrastructure, applications, data, and processes to scale reliably. We design systems with built-in observability, which includes usage monitoring of applications, so we can continuously debug, understand, and improve them. Our scientific stakeholders rely on our solutions and data, making it essential to build and maintain their trust - critical for the ultimate outcome: delivering medicines faster to patients.

Automation enables scale without proportional effort and reduces operational risk. Observability ensures we can detect, diagnose, and improve continuously.

We collaborate, communicate transparently and share knowledge, designs, code, and decisions to achieve collective outcomes and foster trust. We will work in the open (e.g. documented designs and code, documented decisions, peer reviews) and ensure artifacts are searchable. This is critical as we create the CS CoE organization, and we bring together different cultures, geographies, ways of working, approaches, and strengths.

Open collaboration accelerates innovation and builds alignment across diverse teams and disciplines.

We embed security, privacy, and compliance into every design — not as an afterthought - in alignment with Roche corporate guidelines. This includes patient data, licensed/purchased data with contractual restrictions, and data subject to geographic or jurisdictional regulations. At the same time, we design access models that enable appropriate, governed sharing.

Protecting our intellectual property and data is essential for sustainable innovation, faster delivery, and regulatory compliance.

We architect solutions with modularity and flexibility, embedding fault tolerance as a core design tenant. In research and early development, our business priorities shift rapidly, data modalities evolve, and workflows and technologies are constantly changing and improving. Our systems must keep pace with this and scale, pivot, or recover without major redesign.

In gRED and pRED, business priorities, technologies, and regulatory landscapes continuously evolve. Systems that cannot adapt quickly or recover gracefully from disruptions slow down innovation, productivity and increase operational risk.

We ensure all systems, applications, pipelines, and processes are designed to enable Findable, Accessible, Interoperable, and Reusable (FAIR) data from creation through consumption. Data is core to everything we do in the CS CoE.

FAIR data accelerates discovery, enables reproducibility, and can power AI. It ensures that data can be trusted, reused, and integrated seamlessly across the ecosystem.