Back to case studies
Pfizer
AWS Infrastructure Automation for Manufacturing Process Tracking
Deployment time
−80%
Release frequency
monthly
many/day
Infra provisioning
days
< 1 hour
Migration downtime
< 2 hours
Pfizer is an American multinational pharmaceutical and biotechnology corporation dedicated to discovering, developing, and manufacturing medicines and vaccines for immunology, oncology, cardiology, endocrinology, and neurology.
What changed
Before
Infrastructure
Azure
no infrastructure as code
Delivery
Long deployments
days
Scaling
Difficult to scale
After
Infrastructure as code
Terraform
AWS
VPC · IAM · RBAC
Pipelines
GitHub Actions
infra + apps
Flux · Kustomize
Runtime
EKS
rolling updates
Neo4j clusters
Observability
Prometheus
Grafana
An Azure estate without infrastructure as code, replaced by a Terraform-defined AWS platform, with the same GitHub Actions path for both infrastructure and application changes.
What we built
01
AWS Infrastructure Automation and Optimization
With many internal projects following a similar structure, the main goal was to create a core infrastructure that could serve multiple project types and act as a template for new ones. Almost everything was deployed to EKS and AWS-managed services.
Key Features
Designed and deployed scalable infrastructure on AWS using Terraform.
Secured infrastructure using VPCs, security groups, and IAM roles.
Implemented role-based access control per team and project.
Set up a CI/CD pipeline for infrastructure changes using GitHub Actions and Terraform.
Set up a CI/CD pipeline for application deployments using GitHub Actions, Flux, and Kustomize.
Configured and deployed enterprise Neo4j clusters.
Dockerized applications for consistent environments across dev, staging, and production.
Deployed Kubernetes for autoscaling and efficient resource management.
Implemented Prometheus and Grafana for real-time monitoring and visualization.
Results
Reduced deployment time by 80%.
Increased deployment frequency from once a month to multiple times a day.
Improved development team productivity by automating routine tasks.
Achieved near-zero downtime with rolling updates in Kubernetes.
Reduced infrastructure provisioning time from days to under one hour.
100% success in disaster recovery testing with automated backups and redundancy.
02
Migrate Infrastructure from Azure to AWS
The platform ran on Microsoft Azure without infrastructure as code, which made deployments long and scaling hard. The client wanted it on AWS, automated with the same infrastructure-as-code practices as its other projects. A one-off migration, and the most complex job of the engagement.
Key Features
Migrated existing infrastructure from Azure to AWS.
Automated resource provisioning using Terraform and GitHub Actions.
Implemented a fully automated CI/CD pipeline to support continuous deployment.
Set up monitoring, alerting, and automated scaling to enhance operational efficiency.
Results
Successfully automated the provisioning and configuration of the entire infrastructure using Terraform and GitHub Actions, eliminating manual intervention.
Migrated all critical workloads from Azure to AWS with less than 2 hours of downtime.
The new CI/CD pipeline reduced deployment time from days to minutes, enabling continuous delivery of new features.
Autoscaling and enhanced monitoring ensured high availability even during traffic spikes.
03
Airflow and Spark for the Manufacturing Tracking System
Pfizer's teams build a system that tracks and visualizes the manufacturing process. We configured the Apache Airflow and Apache Spark it runs on, taught the team to work with both, and still fix or build smaller parts of its code when needed.
Key Features
Configured Apache Airflow to schedule and monitor the system's data pipelines.
Configured Apache Spark for its data processing.
Trained the client's engineers to build and run their own Airflow and Spark jobs.
Still fix and build smaller parts of the system's code when needed.
Stack
Engagement: Ongoing · Platform & Cloud Engineering, Product Engineering
Next step
Let's look at your platform together.
45 minutes with a senior engineer: your setup, the risks, what to fix first. No sales team.