Your back-end seat comes down to one thing: data models and decisions that still hold up after the tenth client is onboarded and event volume has tripled. That’s the part I care about most — not whether it runs today, but why it will or won’t scale.
I build production Rails on PostgreSQL, and like you, I build with Claude Code in a spec-driven workflow. I own the architecture, data model, and indexing; Claude executes against the spec. My raw-coding background is what lets me catch it the moment it heads somewhere that won’t hold.
Work that maps to what you described:
- Aqualytix — a production Rails + PostgreSQL monitoring platform I built end-to-end with Claude Code: full domain modeling, role-based access, background alerting, deployed on AWS (EC2/S3/CloudWatch/SNS).
- Kuryente Watch — Rails 8 + PostgreSQL using JSONB, with an append-only event log that fans out to live map updates and background push jobs.
- DOST/ASTI — scale-focused backend work: query optimization via root-cause analysis, Redis caching, a message broker for high volume, and k6 load testing before production.
Here’s the shape I reach for when a system has to scale across many clients:
SourceClient API eventsmany tenants, high volume
→
QueueSidekiq + Redisabsorb spikes, retry safely
→
StorePostgreSQL + JSONBindexed, append-only log
→
ServePer-client APIsclean reads, no N+1
Want a short call where I sketch how I’d model one of your client systems?