Wilfried DZADOU
Software Engineer - Backend, DevOps & Infrastructure
New to Contra
Wilfried is building their profile!
Provisioning Ephemeral Cloud Build Runners with Terraform.
Experimented with on-demand Hetzner infrastructure to move expensive Next.js builds from a slow local machine to disposable cloud compute.
I was bored and I wanted to explore a simple question : Instead of permanently paying for powerful build infrastructure, could I provision compute only when I actually need it?
I built a Terraform-based workflow that provisions a temporary Hetzner machine on demand and uses it as a remote Next.js build environment.
Workflow :
- Provision the remote runner with Terraform
- Synchronize the project using rsync
- Execute the production build remotely
- Retrieve the generated artifact
- Destroy the temporary infrastructure
For the experiment, this reduced a roughly 15-minute local build to around 3–4 minutes.
What I explored :
- Infrastructure as Code
- Disposable infrastructure
- Remote build execution
- Compute/cost trade-offs
- Terraform and Hetzner Cloud
This is also a nice demonstration of how I approach infrastructure problems: measure the bottleneck, isolate the expensive workload and test whether infrastructure can remove the constraint.
Next time, I'll add Ansible atop.
0
9
Redesigned a production deployment workflow to reduce application downtime from 3–4 minutes to approximately 20 seconds.
The problem : A production application experienced approximately 3–4 minutes of downtime during deployments because build and release preparation happened as part of the live deployment process.
That meant users were affected by operations that did not actually require the production application to be unavailable.
The approach : I redesigned the deployment lifecycle around a build-and-promote strategy.
Instead of preparing the release directly against the live application:
- application builds happen before the production switch
- release artifacts are prepared independently
- production continues serving the previous version during preparation
- downtime is limited to the final promotion and application restart
The result : production deployment downtime dropped from approximately 3–4 minutes → ~20 seconds (acceptable for the target users)
The change reduced user-facing disruption without requiring a significantly more complex infrastructure architecture.
0
18
Engineering a Multi-Tenant B2B SaaS from Product to Infrastructure.
Designed and built a production B2B SaaS end-to-end, covering backend architecture, multi-tenancy, infrastructure, security, integrations and mobile expansion.
The project : CoutureCiel is a multi-tenant B2B SaaS platform built for tailoring workshops to manage their customers, measurements, orders, inventory, finances and day-to-day operations.
As the sole engineer and CTO, I designed and built the system across the full engineering lifecycle — from product architecture and data modeling to production infrastructure and operations.
- Designed the multi-tenant application and data architecture
- Built core business workflows for orders, measurements, inventory, accounting and customer management
- Implemented granular roles, permissions and subscription quota enforcement
- Integrated WhatsApp Cloud API and transactional SMS for operational notifications
- Designed CI/CD and production deployment workflows
- Built and operated the self-hosted infrastructure
- Implemented monitoring, incident response and infrastructure cost optimization
- Designed a Zero Trust model for sensitive internal administration
- Extended the existing platform architecture toward React Native / Expo mobile applications
My role : Architecture • Backend Engineering • DevOps • Infrastructure • Security • Integrations • Technical Product Engineering
Stack : NestJS • Next.js • Docker • Dokploy • Cloudflare • WhatsApp Cloud API
1
27