How AstraZeneca Accelerates Drug Discovery with GitHub Copilot and Actions

AstraZeneca develops and manufactures prescription medicines for oncology, cardiovascular, respiratory, and rare disease patients globally. Its engineering and data science workforce of more than 5,000 people includes not only software engineers but also computational scientists and researchers who write code as part of drug discovery pipelines. For a company whose core mission is measured in years — a single drug can take over a decade from concept to patient — any technology that accelerates the engineering layer has compounding effects on the business.

Before moving to GitHub, AstraZeneca’s software teams operated across a fragmented mix of tools including Bitbucket and various satellite systems. With roughly 3,000 people accessing source code, collaboration was difficult and deployments were inconsistent. CI/CD pipelines varied by team, making it hard to apply security standards uniformly or spot defects early. The company recognized an opportunity to standardize and accelerate, but the migration itself was high-stakes: 20,000 repositories and 3,000 users couldn’t afford a disruption to business-critical work like clinical trial systems or AI-powered drug discovery platforms.

AstraZeneca approached the migration with a full year of planning followed by a four-month execution phase. The team used GitHub Issues and Actions for the technical repository moves and built custom software to automate the human-side logistics — scheduling, outreach, and permission recreation — for every team. Nearly 99% of the company’s developers and source code were consolidated onto GitHub by the end. The success drew another 2,000 developers and data scientists to adopt the platform organically within the same year, bringing total adoption to 5,000 seats. The team then built a library of reusable GitHub Actions that embedded critical processes — security scanning, automated testing, artifact creation — into every workflow as standard building blocks, doubling the volume of automated CI/CD events within a year.

GitHub Copilot came next. Adoption in a highly regulated pharmaceutical environment required careful governance, and AstraZeneca waited for GitHub’s enterprise legal indemnification before proceeding — a prerequisite to protect the company’s intellectual property. Once that was in place, early results from the Copilot pilot program showed a 40% increase in developer velocity and an estimated 9 to 10 additional hours of output per developer each week. Two-thirds of active users reported that Copilot meaningfully reduced the time they spent on repetitive coding tasks, freeing them to focus on the complex work that only they can do.

For AstraZeneca, the stakes of these gains are unusually high. Every week shaved off the development of a cancer therapy or clinical data platform means that therapy reaches patients sooner. The engineering organization now operates as an active accelerant for the company’s core scientific mission — with automated quality checks catching defects in hours rather than weeks, security embedded by default, and AI assistance compressing the time between an idea and a working implementation.