Systems engineering for things that have to run themselves.
We build software that keeps running without you watching it.
The stack we build and ship on, across every discipline.
Five disciplines, one throughline.
Each one has to keep running without someone watching it.
Built on the stack you already use.
Pick a technology to see what we ship with it, and how.
1import { api, auth, db, trace } from "@burneh/core";23// A typed endpoint: authenticate, query, respond.4export const getOrders = api.route(async (req) => {5 const user = await auth.require(req);67 const orders = await db.orders.findMany({8 where: { userId: user.id, status: "open" },9 take: 50,10 });1112 return trace("orders.listed", { count: orders.length, data: orders });13});You can watch it run.
Everything we ship keeps its own log. On the left, a trading loop taking a signal to a filled order. On the right, a support bot that steps aside the moment a person is needed. The clock is yours, live.
Traced, line by line
A signal lands, the risk check runs, the order goes out. Every line is a real event with a time, one you could pull up a week later. Press Send to fire one yourself.
from @omar on WhatsApp with intent dispute.open
handed to a person, thread attached
from @maria on Telegram with intent booking.create
handled end to end by the agent
It knows when to step back
The bot handles what it can and hands off the rest. The second it is unsure, the whole thread goes to a person, nothing dropped.
The dashboard your operators actually use.
Multi-tenant from the first commit. Live metrics, a full audit trail, and the controls to step in the moment a system needs you.
Overview
all systems · last 60m
It has to be run, not just launched.
So we build the operating side too. Who can touch what, and how the whole thing is holding up, without leaving the dashboard.
Access and roles
Multi-tenant from day one. Every action is scoped to a workspace and a role, with an audit trail behind it.
Learn moreHealth and alerts
Uptime, latency, and error budgets in view, with an alert the moment something drifts.
Learn moreThe code, and what it makes.
No slideware. The thing that ships is the thing you see, side by side with how it is built.
1import { feed, strategy, backtest } from "@burneh/core";23// A mean-reversion strategy, backtested on a year of candles.4const strat = strategy.define({5 entry: (c) => c.rsi < 30,6 exit: (c) => c.rsi > 55,7 risk: { maxDrawdown: 0.05, size: 0.25 },8});910export const report = backtest(strat, feed.history("ES", "y"));The engineering underneath.
The parts you only notice when they are missing. We build them in from the start.
Multi-tenant isolation
Every workspace is sealed off. Data is scoped with row-level security, so one tenant can never read another.
Everything is traced
Every step a system takes is a recorded event with an id, so you can replay exactly what happened.
Rate-limited and abuse-aware
Per-key limits, backoff, and suppression, so one busy or hostile caller can't take the whole thing down.
Safe to retry
Webhooks and jobs carry idempotency keys. A duplicate delivery does the work once, never twice.
Keys stay secret
Secrets live server-side, never in the browser. Every request is HMAC-verified before it runs.
Risk guards on money
Anything that moves funds passes a risk check first, with drawdown and position limits it can't skip.
Sandboxed demos
Every tool can be tried live in an isolated sandbox, so a demo can never touch real funds or data.
Typed end to end
One type system from the database to the UI. If it compiles, the shapes line up.
Ships on your stack
We build in the language and framework you already run, not a rewrite you will have to maintain.
Tell us what has to keep running.
Tell us what it has to do. We come back with a scope, a plan, and where a demo proves it.