The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
🔧 Case study: Rebuilding a live platform's backend architecture — while it's still running.
EarnDLT is a carbon-compliance platform helping companies handle EU CBAM (Carbon Border Adjustment Mechanism) and Scope 3 emissions reporting. When they needed to bring on third-party data providers, the existing GraphQL-only backend wasn't going to cut it — most external integrations expect REST.
My job: migrate a live NestJS GraphQL backend to a REST API architecture, without disrupting the product for existing users.
What that involved:
Mapping existing GraphQL resolvers to REST endpoints without losing functionality
Designing the new API structure for third-party provider compatibility
Keeping the PostgreSQL data layer stable through the transition
Coordinating the migration in a way that didn't require downtime
This is the kind of problem startups run into once a product is live and growing — the architecture that got you to launch isn't always the architecture that gets you to the next stage. Being able to re-architect a running system, not just build new ones from scratch, is a different skill — and it's one I've had to use more than once.
Tech: NestJS · GraphQL · REST · TypeScript · Next.js · PostgreSQL
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Challenges
View all
Related posts
AI Agent Orchestrator decisions determine whether a useful assistant stays useful after day one. For MeetClaw, I design which channels it inhabits, which model provider it uses, and where its state lives before the installation begins. I have organised 3 self-funded buildathons around OpenClaw, Claude Code and Hermes, so I have seen the failures that a topology has to prevent. I work with approvals, read-back and live integrations so the operator can inspect the system. Message me with the workflow that needs coordination.
AI Agent Developer work is most useful when context survives the handoff. On Tenderseal, I structured the flow around reading source material, asking focused questions, and drafting from that context. The product has supported 15+ European companies, while the workflow stays traceable back to the material. I build real integrations with approval points and read-back rather than hiding the hard step behind a chat box. Message me if a document workflow is losing context between people and systems.
Recently worked on my personal project, Digital banking system using Java/Springboot as my tech-stack with a variety of modern enterprise application features namely stateless jwt authentication, pagination, redis caching, role based access control (rbac), etc
