Me postulé a través de un reclutador. El proceso tomó 5 días. Acudí a una entrevista en Tipalti (Tel Aviv) en feb 2023
Entrevista
Interview with requiter, interview with team lead, home assignment, interview with group leaser, hr, phone call with previous colleagues for recommendations, offer. My interview with group leader was canceled as the position has been closed.
Preguntas de entrevista [4]
Pregunta 1
Bellow is a table called fact_daily_users, which contains the users who were active on a specific date with the specific action they have made (aggregate total actions per row). user_id | date | action| total_actions| day_in_row 332 |17/06| view | 1 | 1 332 |20/06| view | 6 | 1 332 |20/06| click | 2 | 1 221 |24/06| view | 4 | 1 221 |24/06| click | 2 | 1 221 |21/06| view | 1 | 2 221 |20/06| view | 1 | 1 332 |21/06| view | 4 | 2 332 |21/06| click | 3 | 2 Q1: write a query that calculates the logic of day_in_row field with SQL, no joins are allowed. "day_in_row" shows consecutive days for user - users that return day after day to the website per row. Q2: write a function (Python / Java) that gets sql query and returns the output in json object
Attached is excel file with 2 tabs: 1: currency exchanges rates to USD per date date |base_currency (USD) | currency | 1 USD | Inv 1 USD 1/1/2020 | USD | Euro | 0.842712 | 1.186645 2: users daily transaction with their currency transaction_id |user_id | value | currency | date 113 | 212 | 4555 | Australian Dollar |1/2/2020 3: users info user_ id | name | location 543 | Hose | MX On a daily basis these excel files are inserted to a S3 bucket (bucket name is given) Q1: Calculate USD value for each transaction by country (SQL / Python)
Q2: Build two collections on MongoDB , or any other No-SQL DB (based on the files from S3) that hold the next information: • Collection 1 that will hold the Mexican users transaction with dollar value and original currency. • Collection 2 that will hold the users who had transactions with more than 2 unique currencies , total transactions value in dollar and last transaction date. Once complete - write a script that import the incremental transactions into S3/CSV or dedicated table on any database (which of you preferred). Hints • You need to set in your code the config for MongoDB ( establish the db , collections etc..) • Think in a way you will be able to use this as a pipeline rather than ad-hoc job
short intro by the manager, then short intro of me, then we went straight to the system design question. the interview went for 90 min and revolved around relational DB
Preguntas de entrevista [1]
Pregunta 1
the question was how to stream data from a relational DB to a cloud data lake in near real-time. The real catch was figuring out how to keep everything transactionally safe - how do you make sure the data lake writes those updates in the exact same order.