Backend — 2024

Clinical Study PlatformPerformance & Security at Norstella

Pharma-intelligence products where the audience is clinical and "it's usually fast" is not an acceptable answer.

Role
Senior Software Engineer
Client
Norstella
Year
2024
Discipline
Backend

The problem

Study pages were slow under high traffic, PatientMatch was surfacing alerts users didn't trust, the Snyk backlog had real XSS and clickjacking findings in it, and the Contentful CMS licence was a recurring cost the team wanted gone.

The approach

Profiling first: I traced the page-load cost to backend query and caching behaviour rather than the front end, and reworked both. For PatientMatch I made the alert logic configurable and tightened the matching rules. On security, I enforced CSP headers and refactored the unsafe code patterns behind the findings instead of suppressing them. Then I led the Contentful → Keystone migration end to end — schema design, data transfer and API integration.

The outcome

Page load times down 55%+ under load, PatientMatch alert accuracy up 25%, 90% of Snyk findings resolved, Sidekiq retries down 30%, and the CMS licence cost removed. Recognised with the company's Gold Star Award.

Results

By the numbers.

Measured, not estimated
55%+
Faster page loads
90%
Security findings closed
25%
Better alert accuracy
Stack

What it's built on.

8 components
Ruby on Rails PostgreSQL Redis Sidekiq Keystone Contentful Snyk CSP
Contact

Let's build
something honest.