Enrolling over a million students into insurance coverage required a week-long blackout period every cycle. The process was manual, the system was effectively down, and errors were common.

We built a Python ETL pipeline connecting the data broker directly to the insurance providers. Enrollment now runs automatically. The blackout period still exists. Now it's used for quality assurance.

The platform had been built and maintained by a small internal team. No branching strategy, manual deployments, no clear path from bug to production. We came in to change that.

We automated the entire deployment pipeline on Azure Pipelines, introduced a structured branching strategy, and gave the team the processes they needed to actually run well. On the product side, we cleaned up the codebase, rebuilt the frontend to be mobile-first, and made it scalable enough to package as a native app using Capacitor. The web and mobile experience now share a single codebase. Time from bug identification to fix in production went from a week to four hours.

The business was evaluating a SaaS product to fill a gap in the integration layer. I proposed something larger: a unified compute architecture. One language across the stack, containerized, cloud-native, and fully automated through IaC. Homogenous by design, so observability, security, and operations could all run from a single layer.

I led the internal effort and built the case across the organization. They approved it and assembled a dedicated team to deliver it.

Members of a major insurance provider had to call in to update their billing details. Every change went through a support agent.

We built a self-service billing system that handles payment updates directly, PCI compliant, fully serverless. Call center volume dropped. So did the overhead.

The existing enrollment and payment pipeline crashed monthly. Total outage costs ran into the millions.

We built a fully automated integration that replaced it. Data flows in, gets processed, and lands in the right place without intervention. The system runs at 99.99% uptime.

Tableau deployments are usually manual, undocumented, and tied to whoever set them up.

We wrote the entire environment as infrastructure as code. One Serverless Framework file deploys everything: EC2 behind a load balancer, private VPC, role-based access controls, and a direct connection to the Redshift data warehouse. The whole setup is version controlled and can be redeployed from scratch. Business teams get real-time access to the data they need without filing a report request. IT gets an environment they can actually maintain.

Monthly close at a regulated financial institution meant weeks of manual effort, no audit trail, and no visibility into whether the underlying data had changed.

We built a reporting engine that generates reports on demand, detects stale or inconsistent data automatically, and logs every modification including post-close changes. Reporting that took a month now takes minutes.