Sobes.tech

Data Engineer

Tell us about yourself: where did you work last, what did you do, what are you looking for, and why?

Middle+
2

Where are Driver and Executors located in cluster mode and why is it used in production?

2

Where exactly does the gradient occur in the gradient boosting algorithm if the base algorithm is a simple decision tree?

Middle+
2

How do you use Python in your work and where do you feel more confident: in Python or SQL?

Senior
2

There are 101 objects: 100 objects of class 0 and 1 object of class 1. The first model gives the same score to all objects. What are its ROC AUC and the shape of the ROC curve?

Middle+
2

How to test a time series for stationarity?

Middle+
2

Given an SQL query written by an analyst. Identify inefficiencies in the query and suggest how to avoid them. The query joins two tables: - fact table 'events' with key 'event_id' - traffic source reference table with key 'referer_str' Both tables contain billions of records. select * from (select distinct e.event_id, e.user_id, e.category_id, e.location_id, s.source_name, s.medium_name, s.campaign_name, row_number() over(partition by e.event_id, e.user_id, e.category_id, e.location_id, s.source_name, s.medium_name, s.campaign_name) as rnk from events e left join traffic_source s on e.referer_str = s.referer_str where e.event_date = current_date() - 1 group by 1,2,3,4,5,6,7) t where rnk = 1

Senior
2

Data has been loaded into a ReplacingMergeTree table; what will happen during reading if the merge has not yet occurred, and how to get the latest version of the records?

2

What is denormalization?

Middle
2

Is it possible to combine deduplication and subsequent data spill without using separate tables; and if so, how?

2

What are index structures and how do they work?

Middle
2

What is a stationary time series and why is stationarity important for classical models?

Middle+
2

What is bagging and boosting, and how do they differ?

Middle+
2

What are the features of using JOIN when building data marts in PostgreSQL/Greenplum compared to ClickHouse?

Middle
2

There is a table t: | id | id_client | create_date | email | |----|-----------|-----------------------|--------------------| | 1 | 1 | [phone]T10:00:00.000Z | [email] | | 2 | 1 | [phone]T11:00:00.000Z | [email] | | 3 | 2 | [phone]T08:00:00.000Z | [email] | | 4 | 2 | [phone]T10:00:00.000Z | [email] | | 5 | 2 | [phone]T11:00:00.000Z | [email] | | 6 | 3 | [phone]T08:00:00.000Z | [email] | | 7 | 3 | [phone]T09:00:00.000Z | [email] | | 8 | 4 | [phone]T12:00:00.000Z | [email] | | 9 | 4 | [phone]T13:00:00.000Z | [email] | | 10 | 5 | [phone]T14:00:00.000Z | [email] | | 11 | 5 | [phone]T15:00:00.000Z | [email] | Нужно проверить, есть ли в таблице клиенты с несколькими email. Если есть, то удалить их (оставить только последнюю запись по create_date).

Middle
1

What is normalization and when is it applied?

Middle
1

In a six-shooter revolver, two bullets are consecutive. The first player pulled the trigger, and no shot was fired. Is it more advantageous for the second player to immediately pull the trigger or to spin the drum? Justify.

Middle+
1
/38