- Sheridan, Wyoming, US
- [email protected]
- Call Us Anytime+123 (4567) 890
- Email Us[email protected]
- Opening HourMon - Sat 8:00 - 6:30, Sunday - CLOSED
Get latest update for our trusted applications
Careers
Build AI that has to pay for itself.
Most AI jobs end at the demo. Here the work is only finished when a business number has moved, which is harder, considerably less glamorous, and the reason our engineers stay an average of four years.
217
peopleAcross seven countries
21
open rolesCurrently hiring
4.2 yrs
average tenureOf delivery staff
You will find out whether your work actually worked.
In most engineering jobs, feedback ends at deployment. The system goes live, the team moves on, and nobody establishes whether any of it mattered. It is a strangely unsatisfying way to spend a career — years of competent work with no way to know if it was useful.
Here every engagement carries a value model, and the number gets reported monthly. You will know that the collections agent you built recovered eighteen per cent more, or that the adoption you designed for came in at 94%, or that the forecast you were confident about turned out to be wrong. All three of those are more interesting than a successful launch.
It also means the work is harder. Production reliability, adoption, integration with genuinely awful legacy systems, and the discipline of holdout measurement are all part of the job. If you want to train models and hand them over, this will frustrate you.
Three structural choices you will feel weekly.
CHOICE 01
No graduate bench, no pyramid
We do not hire juniors to be billed as seniors. Engineers are hired to own work, and the person who scopes an engagement builds it. Career progression is depth of ownership rather than distance from the code.
CHOICE 02
You will be in front of clients
There is no account layer to hide behind. Engineers present in value reviews and hear directly when something is not working. Uncomfortable at first, and the fastest professional development available.
CHOICE 03
Refusing work is respected
Saying that a proposed approach will not create value is a contribution, not an obstruction. Several of our most valued internal moments have been engineers arguing successfully against a build.
Twenty-one positions, mostly in Kathmandu.
Engineering is centralised deliberately. Client value and design roles have more geographic flexibility.
Senior AI Engineer — Agentic Systems
Kathmandu
Machine Learning Engineer — Decision Intelligence
Kathmandu
Speech & Language Engineer — Voice AI
Kathmandu
Senior Full-Stack Engineer — TypeScript
Kathmandu
Data Engineer — Warehouse & Pipelines
Kathmandu
Site Reliability Engineer
Kathmandu
Product Designer — Operational Interfaces
Kathmandu / remote
Value Consultant — Assessments
Amsterdam / London
Client Partner — Middle East
Dubai
No role that fits? Write to [email protected] with something you have built and the business outcome it produced. That second part is what we read first.
Four steps, no puzzles.
We do not ask candidates to invert binary trees or complete unpaid work. The process is designed to find out whether you can build something that works and explain why it mattered.
Conversation
Forty-five minutes with a practice lead about what you have built and what it achieved. No screening call with a recruiter who cannot answer technical questions.
Technical discussion
A deep session on your own past work: decisions, trade-offs, what went wrong and what you would change. Bring real systems, not prepared answers.
Practical exercise
A short, paid, realistic problem resembling actual client work. Reviewed by the engineers you would work with, with written feedback either way.
Decision
An offer or a clear no within a week, with the reasoning. We do not leave candidates waiting, and we do not disappear.
What we offer, and what we expect.
- 01Compensation at the top of the local market, reviewed twice yearly
- 02Ownership of real systems in production, not tickets
- 03Direct client exposure from your first quarter
- 04A funded learning budget and conference time
- 05Health cover extending to family
- 06Group backing, meaning stability through downturns
- 01Willingness to be measured on outcomes, not activity
- 02Comfort saying no to work that will not create value
- 03Care about adoption, not only about correctness
- 04Directness in status reporting, especially bad news
- 05Patience with legacy systems and imperfect data
- 06Interest in the client's business, not only the technology
Who you would join.
The numbers we watch on ourselves.
4.2 yrs
tenureAverage across delivery staff
61%
senior+Share at senior level or above
9%
attritionAnnual voluntary attrition
34%
internal movesStaff who changed practice internally
- 01Promotion decisions reviewed across practices, not by one manager
- 02Internal moves between practices actively supported
- 03No obligation to move into management to progress
21
Open rolesAcross five practices
4.2 yrs
Average tenureOf delivery staff
9%
Voluntary attritionAnnual, across the firm
What candidates ask.
- 01. Is remote work possible?For selected design and client value roles, yes. Engineering is centralised in Kathmandu because that concentration is what produces our consistency, and we would rather be honest about that than advertise flexibility we do not offer.
- 02. Do I need AI experience to join the AI practice?You need strong engineering fundamentals and evidence that you can take something into production. Several of our best AI engineers came from backend and data engineering backgrounds and learned the modelling here.
- 03. What does progression look like?Depth of ownership: from contributing to a system, to owning one, to setting the standard for a capability. Management is one path among several and not a requirement for seniority or pay.
- 04. Is the practical exercise paid?Yes, at a fair rate for the time involved, and you receive written feedback whatever the outcome. We do not ask candidates to produce unpaid work.
Where to go next.
Tell us what you built, and what it was worth.
Write to [email protected] with the system and the business outcome. We read the second part first, and we reply either way.
