What we test for
Typical work
- A warehouse the business can query without asking which number is right
- Pipelines that survive a late file, a duplicate and a backfill
- Moving reporting off the production database before it takes the product down
- The data layer under a product feature: events, aggregates, and what they cost to keep
The coding test
Take a feed that arrives late, sometimes twice, and occasionally with yesterday corrected, and load it so the same query gives the same answer on every run. Then we replay three days out of order and ask for the daily totals. Scored on idempotency and on what they chose to store rather than recompute.
Inside Data engineering
The ones people search for by name get a page of their own, with their own coding test and interview questions.
dbt and analytics engineering
Models with tests and a lineage you can read, reporting numbers that match the ones finance already published.
See the page →Airflow and orchestration
Pipelines that recover on their own: idempotent tasks, backfills that do not duplicate, and alerts that mean something.
See the page →Five interview questions you can reuse
We use questions like these in our soft-skills and technical rounds. Steal them for your own interviews.
Two dashboards give different numbers for the same metric. How do you settle it?
A strong answer covers: Go to the definition before the data. Most of the time nothing is broken: one counts orders placed and the other orders paid, or they cut the day in different time zones, or one filters cancellations and the other does not. The fix is one defined metric in one place that both read, not a reconciliation done by hand every month.
What makes a pipeline safe to re-run?
A strong answer covers: That running it twice leaves the same state as running it once. In practice that means loading a partition by replacing it rather than appending, keying on something stable from the source instead of on arrival order, and keeping the transformation a pure function of its input. Anything that reads the current time or the current row count during the load will not survive a backfill.
When is Spark the wrong answer?
A strong answer covers: Most of the time, now. A warehouse will do a hundred gigabytes without anyone managing a cluster, and the operational cost of Spark is real. It earns its place on genuinely large data, on work that is not expressible in SQL, or on files nobody wants to land first. Reaching for it by default is how a team ends up with two engines and one of them untested.
How do you keep slowly changing dimensions honest?
A strong answer covers: Decide, per column, whether history matters, and write that decision down. Overwriting is right for a corrected typo and wrong for a price. Where history matters, keep the row with its valid-from and valid-to and join on the date of the fact, not on the key alone. The classic bug is last year's revenue moving because a customer changed their region.
What do you do about a pipeline that is correct but expensive?
A strong answer covers: Find out what is actually being paid for first: usually a full scan where a partition would do, or a model rebuilt hourly that nobody reads before Monday. Incremental models, partition pruning and a cadence that matches the question are worth more than tuning. And some tables are cheaper to drop than to keep.
One invoice a month.
The shortlist is free. Pick the contract that suits you: monthly, six months or twelve.
What every seat includes
- Contracts, payroll and compliance handled
- Vetting: coding test, soft skills, English and a level check
- Free replacements: unlimited on 12+ months, 2 on 6 months, 1 on monthly
- A direct line to our CEO
Terms in plain words
- Monthly, six months or twelve, your choice
- Your rate is locked for the whole contract on 6 and 12+ months
- Monthly ends with 30 days’ notice
- Want to hire the developer full-time? The buyout is set in the contract from day one
Questions tech leads ask us
How senior are your Data engineering engineers?
6+ years of professional Data engineering work, with recent production experience.
Can I see the coding test results?
Yes. Every profile card includes the score and the reviewer's notes.
Skip the sales rep. Talk to the CEO.
Tell us what you're building, your stack and when you need people. Three vetted profiles per role within 48 hours of the call, and a straight answer on whether we're the right fit.