Failures stay small
When one part crashes, it restarts on its own in milliseconds. The rest of the system keeps serving customers.
02 Backend systems
The part your users never see, and always feel.
- Elixir
- Phoenix
- Node
- GraphQL
- Postgres
- Cloudflare
The engine room decides whether your busiest day is a good day or an incident. It is also the part nobody sells you properly — so it is usually the part that fails.
Start a projectWhy it falls over exactly when it matters
- 01 Traffic doubles and the whole system goes down together
- 02 A payment fails halfway and the order simply disappears
- 03 One slow query takes down every page at once
- 04 Business rules live in six places and disagree with each other
↓
You need failure contained instead of total, your business rules written down once, and a backend chosen for the load you actually expect — not for whatever the last team happened to know.
What you get
Nothing gets silently lost
Background work is stored with your data in the same transaction. A crash mid-payment means the job retries — not that it vanished.
Your rules, in one place
Permissions and validations declared once in the domain layer, so they cannot be forgotten in the one screen nobody tested.
A stack sized to your project
Elixir where concurrency and uptime genuinely matter, something simpler where they do not. The delivery network in front and the database underneath stay the same either way.
How we build it
- How we deliver it
- Elixir, when it earns its place
- For systems with thousands of simultaneous users, live updates or long-running connections, Elixir removes an entire category of difficulty. For a straightforward CRUD API it is overkill, and we will tell you so.
- Supervised processes
- When a component crashes, something restarts it in milliseconds rather than the whole system going down with it. One bug becomes a blip instead of an outage.
- Ash Framework
- Your business rules become a declaration rather than code scattered across the app — which is what makes them auditable.
- Oban
- Background jobs stored in Postgres inside your transaction, with retries and history you can inspect when a customer asks what happened.
- Postgres, modelled properly
- Indexes and access patterns designed against your real queries, so growth does not turn into a rewrite.
When to bring us in
Every project starts with a scoping call- ↳ Your system goes down on your busiest days
- ↳ You handle payments, bookings, or anything you cannot afford to lose
- ↳ You need real-time — live updates, presence, notifications
- ↳ Your database is the bottleneck and nobody knows why
- ↳ You are planning for ten times the users you have now
- ↳ You need an honest architecture review before committing budget
What we build
Services- 01 Web applications You do not need another redesign. You need an interface built so that adding to it is safe and speed is not something that degrades over time. SvelteKitNext.jsTypeScript
- 03 Mobile applications You need one codebase that treats both platforms as first class, works offline, and is still buildable in two years. React NativeExpoTypeScript
- 04 Custom software The work already has a process. Software should encode that process, remove the retyping, and let you see what is happening without asking someone. ElixirAshTypeScript
- 05 Cloud & delivery Infrastructure should be written down as code, reproducible from scratch, and monitored — so recovery is a procedure rather than a rescue. CloudflareTerraformRailway
Build it to hold.
Describe the load you expect and what you cannot afford to lose. We will tell you what the system needs to look like.
Start a project hello@elixiria.ma