What we test for
Typical work
- Serving models at a latency and a cost the product can actually afford
- Pipelines that retrain and redeploy without a person in the loop
- Versioning models, data and prompts together, so a rollback is possible
- Monitoring for drift, and the alerting that has to go with it
The coding test
Take a model and a notebook and make it a service: a reproducible build, a versioned artifact, a deploy that can roll back, and the one metric you would alert on. Then halve the cost per request and tell us what you traded away. Scored on reproducibility and on what they chose to measure.
Five interview questions you can reuse
We use questions like these in our soft-skills and technical rounds. Steal them for your own interviews.
What makes a model deployment reproducible?
A strong answer covers: Code, model weights, data snapshot and config versioned together and built in a pipeline rather than on a laptop. If you cannot rebuild the deployment from last month exactly, you cannot roll back to it either.
How do you cut inference cost without wrecking quality?
A strong answer covers: Measure where it goes first. Then batching, quantisation, a smaller or distilled model on the easy path, caching, and right-sizing the GPU. Each one is a trade you should be able to quantify before you make it.
What do you monitor on a model in production?
A strong answer covers: Input drift, output distribution, and the business metric the model is meant to move, plus latency and cost. Accuracy usually arrives weeks late or never, so the leading indicators are what you actually alert on.
The model, the data and the prompt all changed and quality dropped. How do you find which?
A strong answer covers: Version all three and be able to pin two while moving one. Without that you are bisecting a system with three moving parts and no record of which one moved.
When is a model not the right answer?
A strong answer covers: When rules would do the job, when there is not enough labelled data to tell whether it works, or when nobody will own it after launch. A model with no owner becomes an unmonitored liability within a quarter.
One monthly price per developer. Everything in it.
No deposit, no minimum term, no placement fee. You pay one invoice a month for each developer on your team.
Per developer, all-in. Your quote arrives by email with your exact number after you send the brief.
- Developer pay, local payroll and taxes
- Vetting: coding test, soft skills, English
- Free replacement within 90 days
- A direct line to our CEO
What changes the price
- Seniority and role. Senior engineer, tech lead or architect.
- Stack. Some stacks and niche tools cost more to staff.
- Team size. Larger teams get a better monthly rate per seat.
- White-label. Agencies reselling under their brand get partner pricing.
Terms in plain words
- First 2 weeks risk-free: if it isn't working, you don't pay for them
- Month to month, 30 days' notice to end
- No deposit and no minimum number of seats
- Want to hire the developer full-time? Buyout of 3 months' fee in year one, 2 months' after
Questions tech leads ask us
How senior are your MLOps engineers?
Six or more years in infrastructure or ML engineering, with models they put into production and then had to keep running. The second half is the part that is hard to fake.
Can I see the coding test results?
Yes. Every profile card carries the score and the reviewer notes.
What does it cost?
$6,900 per developer per month, all-in, and flat: the same rate whatever the stack. Your exact quote arrives by email after you send the brief.
Skip the sales rep. Talk to the CEO.
Tell us what you're building, your stack and when you need people. Three matched profiles within 48 hours, and a straight answer on whether we're the right fit.