Nexus

Engineering practice

Software that still works when nobody is watching it.

Nexus designs, builds and operates systems that run unattended. Fixed scope, fixed price, and evidence before claims — starting with the infrastructure we run ourselves.

Start a conversation See the proof

Every engagement is scoped in writing before a line of code is billed.

Recognition

You already know the shape of the problem.

These are the conditions we are usually called into. If none of them are familiar, we are probably not the right people.

  • It holds while someone is watching

    The system behaves during working hours. The failures arrive at night and on weekends, and nobody finds out until someone complains.

  • Nobody can say what it does

    The behaviour lives in the heads of two people and in code nobody has read end to end. Every question becomes an investigation.

  • Each change costs more than the last

    Work that took two days last year takes two weeks now. The estimate is no longer a forecast — it is a negotiation.

  • The numbers cannot be checked

    Dashboards report success while the underlying job has been failing quietly. Nobody trusts the reporting, so nobody uses it.

Method

Four steps, in this order, every time.

The sequence does not change per client. What changes is how long each step takes and what it turns up.

  1. Read the system

    Before anything is written, we map what already exists: the code paths that run in production, the jobs nobody remembers scheduling, the credentials, and the failure history. The output is a written account of the system as it actually is.

  2. Fix the scope

    One page. What is being built, what is explicitly excluded, what it costs, and what has to be true for it to be finished. Nothing starts until both sides have signed the same page.

  3. Build in the open

    Working software every week, in your repository, on your infrastructure, with the commit history intact. Progress is something you can run, not something you are told about.

  4. Hand over cleanly

    Runbooks, architecture notes, monitoring, and a recorded walkthrough. The engagement ends when your team can operate the system without us — which is the only definition of finished we accept.

Proof

The first proof case is our own infrastructure.

We do not open with client logos. We open with a system we built, run and pay for ourselves — and publish what it does, including what it has not done yet.

Nexus autonomous content infrastructure

  • Built and operated in-house
  • Running continuously
  • Single server

An unattended pipeline that takes a topic from keyword research through drafting, editorial gating, on-page SEO, media generation and publication to a live production site — then reports on what it published. It runs on one virtual machine under a process supervisor, recovers from provider rate limits on its own, and quarantines work that falls outside its subject boundary rather than publishing it.

Pipeline stages
7
Articles published
43
Manual steps per publication
0
Servers
1

It is a working system, not a success story. Distribution is the open problem: the archive is published but under-indexed, and the operating notes say so in the same detail as everything else. That is the standard we hold client work to as well.

Read the full case

Scope

What we take on, and what we decline.

Declining the wrong work early is the cheapest thing either of us can do.

We take on

  • Systems that must run unattended: pipelines, schedulers, integrations, background workers.
  • Production hardening — monitoring, recovery, rate-limit handling, failure reporting.
  • Rescue of systems whose original authors have left.
  • Internal tools that a business already depends on and cannot afford to lose.
  • Second opinions on architecture, delivered in writing.

We decline

  • Hourly staff augmentation and open-ended retainers.
  • Discovery phases with no defined end.
  • Work where success cannot be measured by anything we can run.
  • Marketing claims about results we have not observed ourselves.
  • Engagements where we would not hold the credentials to verify our own work.

Terms

The terms are the same for everyone.

They are published here so that the first conversation can be about the problem instead of the contract.

Engagement
Fixed scope at a fixed price, agreed in writing before work starts.
Cadence
Working software every week, in your repository, with a written summary of what changed.
Ownership
Code, infrastructure definitions and documentation are yours from the first commit.
Access
You hold the credentials throughout. We work in your accounts, never around them.
Exit
Stop at the end of any week. No notice period, no termination fee, handover included.
Confidentiality
Mutual NDA on request. Your name is never used as a reference without written consent.

Send the problem, not a brief.

Describe what breaks, how you find out, and what it costs when it does. You get a written assessment in return: scope, price, and an honest answer on whether we are the right people for it.

hello@nexus.studio