Instinctools Data Engineer Interview Experience
Instinctools · Senior
Interview process
Overall: The interview was well structured and closely matched the role. It covered SQL and BigQuery performance, dbt pipelines, data modelling, Airflow, data quality, Python, debugging, cost, and cloud platforms. The scenarios were realistic for an attribution-pipeline migration, which made it a good test of practical skills rather than theory.
What went well:
The questions were scenario-based and relevant: late-arriving attribution data, restated ad-network spend, backfills, and duplicate investigation.
The interviewer followed up on my answers and rephrased or repeated questions when I asked, so I could go deeper.
The pause option and the freedom to take time to think were helpful.
I felt strongest on dbt pipeline design, data quality testing and investigation, and data modelling for marketing data.
What didn’t go well:
In the data quality section, the screen kept showing questions from earlier sections while new ones were asked by voice, and the check-in question was repeated several times. This broke my concentration for a while.
The suggestion to turn away from the screen was confusing, because the questions were displayed on that same screen.
The session ran about 10 minutes over the scheduled time.
On my side, a few BigQuery-specific questions, such as partition pruning behaviour and query plan metrics, deserved sharper answers. I’ve since reviewed those topics.
There was no opportunity to ask about the team or the project, which I’d normally do with a technical lead.
Interview rounds · 4
- 1
Recruiter screen
Q1. Tell me about your experience.
How they answeredmy experience in pharma, consulting, startup
- 2
Online assessment
Q1. How do you collaborate with engineers?
How they answeredrequirements gathering, workshops
- 3
Phone screen
Q1. Tell me about your biggest failure.
How they answeredwith project scope
- 4
Technical round
Q1. What's the difference between Bigquery and Snowflake?
How they answereddifferences between bigquery and snowflake
Tips from the candidate
Know BigQuery internals well: partition pruning (why it fails when the date filter comes from a join), reading the query execution details, INFORMATION_SCHEMA.JOBS, job labels, and how to repartition an existing table.
Understand idempotency: use Airflow’s logical date instead of the current date, make reruns and backfills safe, and choose between partition overwrite and MERGE.
Know dbt incremental models in depth: strategies on BigQuery, lookback windows for late data, targeted backfills with vars, and keeping invalid rows out of marts without stopping the run.
Review marketing data modelling: late-arriving dimensions, multi-touch attribution with weights, and handling restated ad-network spend.
Be precise on SQL fundamentals: WHERE vs HAVING, join fan-out, and what makes a query non-deterministic.
Prepare basic Python engineering: rate limiting and retries for APIs, and sharing code as a versioned package.
Company culture
I haven’t met the engineering team yet, so my impression comes from the hiring process so far. A few things stood out:
Transparency. The job posting was unusually clear about the project, the interview stages, the contract model, and how payment works. That made it easy to judge whether the role was a good fit before applying.
Speed and responsiveness. The recruiter reached out quickly, the first call was well organised, and the next step was scheduled without delays.
High technical standards. The technical interview was demanding and closely tied to the real project work: attribution pipelines, BigQuery, dbt, and migrations. That tells me the company places people carefully rather than just filling seats.
A small, personal company. Being a family-owned team in Lisbon that builds engineering teams for clients suggests a more personal relationship with contractors than at a large outsourcing firm.