Microsoft Data Science interview questions

Do you have techniques that you can you illustrate to scoop the data from the set of source tables into the Cube? What is a main calculation and how you use it in a Cube?

Explain the steps to log data into Azure SQL pool from blob storage TheHolidays?

Can you explain what is a Virtual Cluster in Azure Cosmos? What is Stream in Azure Cosmos?

Azure COSMOS Job submision - how to prioritize the execution for a job? How you debug a failed job? How you do a performance tuning?

SCOPE scripting: There are 2 daily structured streams stored in COSMOS, you need to extract those for the last 30 days. 1st - customer usage SS (i.e. this customer used x,y,z servises and total usage is A), 2nd- customer activity events for particular service (what customer is actualy executing, from the services standpoint, for example customer provisioned an sql db, total usage is 1000s per day, so what is he doing from activity standpoint: running queryes, uploading data, etc.) No real sructure provided, need to show how you think about this problem. Hints: 1st SS might have a date, numerical usage number, customer_id, product_id. Expect to write a script to extract all related SS files, and queries to produce the result.

Explain all types of joins in SQL (using whiteboard, with pictures ). Having OrgTable(guid,name,office_loc,headcount,org_guid) given org_guid has 1-M relationship with guid, write a query to return team_guid,team_name,org_name,org_guid (to show which org each team belongs to) by leveraging a CTE (Common Table value Expression) recursive query (I guess self join).

Explain the 5 key (The 5 Vs) factors to consider when building Big Data pipelines: Volume, Velocity,Variety,Veracity(data quality) and Value. Explain techniques to check for Veracity?

Explain data structure techniques that can be used to determine an existence of the data element (what to use to determine if data element exists) in Big data model? Hint: a lot of structured and unstructured data coming in, so how would I check if some data exists in particular row? Like bunch of telimetry data streaming in, think of row of a multiple data values, how to check if one specific value already exists in that row? Need to tell a data structure that you use (or design) to do it.

In R generate 1000 data points normally distributed, with mean=x, and variance=y. Perform summary stat on the data points, explain stat metrics. Create density plot. Do the same in python scrypt, in codility.com editor

Explain right-skew and left-skew distributions (draw on whiteboard). Where are mean and median values in these cases, compared to symmetrical normal bell shape?