Topher Warrington

About the Author

Welcome to Topher Codes! I’m Topher, a passionate web designer, coder, and project manager. With over a decade of experience in the tech industry, I share my insights, tips, and personal musings on this platform. Dive in to explore my journey and learn from my experiences.

Post By Topher

AI Content Moderation: Keeping Prayer Communities Safe

AI Content Moderation: Keeping Prayer Communities Safe

by | Apr 28, 2025 | ,

Keeping online prayer communities safe means walking a fine line between compassion and security—especially as your platform grows. In this post, I share how we implemented AI-powered content moderation using Claude Sonnet 4.5 to help Prayer Nook handle 200+ daily requests. Learn why human-in-the-loop review is essential, how we engineered Rails service objects for moderation, the real results (including urgent crisis detection), and the ethical guardrails we put in place. If you’re considering AI for community safety, don’t miss these production-tested lessons—and a look at the future of faith-tech.

Building a Custom SSO: Why We Chose OAuth2 Over Devise

Building a Custom SSO: Why We Chose OAuth2 Over Devise

by | Mar 31, 2025 | ,

In 2021, we made a decision that seemed crazy: build our own OAuth2 Single Sign-On system instead of using Devise. Three years and three Rails upgrades later, that “crazy” decision looks like strategic foresight.

This is the story of Heis Soma—our custom authentication service powering Prayer Nook and positioning us to serve the broader Christian ministry ecosystem. It’s about architectural decisions, technical tradeoffs, and the surprising ways that following standards can future-proof your applications.

We built ~2,000 lines of OAuth2-compliant code. Survived Rails 6.1 → 8.0 with minimal changes. Achieved 10x performance improvements through Rails 7’s multiple database connections. Served 1,000+ users with zero security incidents.

Was it worth the 200-hour investment? Yes—but not for everyone. Here’s when custom authentication makes sense, when it doesn’t, and what we learned from three years in production.