Careers
Small team, senior only, no layers
We hire slowly and rarely. When we do, it is because a client engagement needs a skill we do not already have — not because a headcount target says so.
What it's like
You will do the work you were hired for
Everyone here is hands-on. The person who scopes an engagement is the person who builds it, and there is nobody in between writing status decks about it. If that sounds obvious, it is worth saying that most agencies do not work this way.
The work is genuinely varied and often unglamorous. A good week might mean deleting a spreadsheet, replacing four manual handoffs with one, and writing the runbook that lets a client's team operate it without us.
We are remote-first across four continents, with offices in Plano and Thiruvananthapuram if you would rather not work from home. Overlap matters more than location: we ask for four hours a day that line up with the rest of your project team.
- No on-call rota. We build systems that alert their owners, not ones that wake us up.
- No billable-hours target. Engagements are fixed-scope, so nobody is optimising a timesheet.
- Your name on the work. Case studies credit the people who built the thing.
Open roles
What we are hiring for
Three roles open right now. If none of them fit but you think we should talk anyway, the last one is for you.
Senior Automation Engineer
You will own client engagements end to end: sitting in the audit, designing the workflow, building it, and handing it over. Most of the work is orchestration — n8n, Make or Temporal — wired into whatever CRM, billing and comms stack the client already pays for.
We are looking for someone who has built integrations that stayed up. Experience with retries, idempotency and error handling matters more here than any particular tool. You should be comfortable talking to an ops lead about their invoicing process without reaching for jargon.
Data Engineer
Half our projects turn out to be data problems wearing an AI costume. You would build the pipelines underneath them: consolidating scattered sources into a warehouse, reconciling the numbers that disagree, and standing up the reporting a founder actually opens each morning.
We are looking for solid SQL, comfort with dbt and either BigQuery, Snowflake or Postgres at scale, and the judgement to say when a client's data is not yet good enough to model on. That last one is most of the job.
AI Solutions Engineer
Retrieval agents, support deflection, document extraction — the LLM-shaped half of what we do. You would build the evaluation harness before the prompt, ship behind confidence thresholds, and stay honest about what the model cannot be trusted with.
We are looking for someone who has taken an LLM feature past the demo. Evals, cost and latency work, and grounded retrieval count for more than prompt-craft. Scepticism about your own output is a requirement, not a red flag.
None of the above
We keep a short list of people we would hire when the right engagement lands, and we do go back to it. Tell us what you have built, what you would rather not do again, and roughly when you might be looking. We answer every one of these, though not always quickly.
How we hire
Four steps, about three weeks
No whiteboard algorithms, and nothing you have to do at midnight after your actual job.
-
01
A reply
Every application gets an answer, including the ones we pass on. If we are moving forward you will hear within a week.
-
02
A conversation
Forty-five minutes with whoever you would be working alongside. Mostly about things you have actually built, and what went wrong in them.
-
03
A real problem
A scoped exercise drawn from work we have genuinely done, capped at three hours. We pay for your time on it, whatever the outcome.
-
04
An offer
A walkthrough of your exercise with the team, then a decision within two days. We share the salary band before you start, not after.
Questions first
Not sure whether to apply?
Ask before you spend an evening on a cover letter. We would rather answer a direct question than have you guess at whether it is worth it.