Amazon · amazon.jobs · checked today
Data Engineer II, Alexa AI - ML Engineering
Join Alexa AI - ML Engineering and help power personalized, proactive experiences for hundreds of millions of Alexa customers worldwide. As a Data Engineer II, you will design and own the large-scale data pipelines that turn raw customer-interaction signals into the context and features our personalization and proactive-agent models depend on, working side-by-side with SDEs to shape our data strategy and architecture. - You will build and operate ETL/ELT pipelines processing billions of records a day, using AWS big-data services (Glue, EMR, Kinesis, Redshift, DynamoDB) and distributed compute.
Skills, with evidence
- Data modelling
Experience with data modeling, warehousing and building ETL pipelines
must have - SQL
4+ years of SQL experience
must have - Warehousing
Experience with data modeling, warehousing and building ETL pipelines
must have - AWS
Experience with AWS technologies like Redshift, S3, AWS Glue, EMR, Kinesis, FireHose, Lambda, and IAM roles and permissions
must have · not practised here - Spark
- You will build and operate ETL/ELT pipelines processing billions of records a day, using AWS big-data services (Glue, EMR, Kinesis, Redshift, DynamoDB) and distributed compute.
- Streaming
- You will build and operate ETL/ELT pipelines processing billions of records a day, using AWS big-data services (Glue, EMR, Kinesis, Redshift, DynamoDB) and distributed compute.
- Cost & performance
Operate and continuously improve production pipelines: optimize data delivery, redesign for scalability, manage deployments, and drive root-cause fixes for pipeline incidents.
- Data quality
We value ownership, data quality, and working backwards from the customer.
- Failure handling
Operate and continuously improve production pipelines: optimize data delivery, redesign for scalability, manage deployments, and drive root-cause fixes for pipeline incidents.
- Schema evolution & contracts
Partner with applied scientists and ML engineers to define data contracts, feature pipelines, and data-quality checks feeding personalization and proactive-agent models.
Your plan
- The SQL screen: correct, then fast≈ 1 h
SQL · Data quality
- Average order value by countryFoundations
- Line items per orderFoundations
- Revenue by buyer countryFoundations
- Customers who never orderedFoundations
- Hiring funnel by roleFoundations
- Data modelling: the round most people fail≈ 2 h
Data modelling · Warehousing
- Marketplace core entitiesFoundations
- Cinema seat bookingFoundations
- City parking baysFoundations
- Dating app matchesFoundations
- Food delivery ordersFoundations
- Pipeline design: safe to run twice≈ 3 h
Streaming · Failure handling · Schema evolution & contracts
- After the first one finishesFoundations
- The column that vanishedFoundations
- The history nobody keptFoundations
- The key that does not matchFoundations
- The spreadsheet is a dependencyFoundations
- Spark: read the plan Spark actually ran≈ 1 h
Spark · Cost & performance
- HAVING vs WHERE: where does a filter after groupBy actually run?Foundations
- countDistinct vs approx_count_distinct: what the extra shuffle buysFoundations
- COUNT(*) vs COUNT(column): the null trapFoundations
- Grouping by two columns: what changes in the shuffle?Foundations
- Does the join type change the join strategy?Foundations
- Say it out loud≈ 1 h
Not covered by the plan: AWS.
Readiness
Counted from drills you have completed anywhere on D8LooP.
leaves in 12 dremoved the moment Amazon closes it