AWS
70 bites tagged AWS — interview questions with model answers, and 60-second explainers.
Explain the Well-Architected Framework pillars
Name the pillars, operational excellence, security, reliability, performance efficiency, cost optimization, and sustainability, and explain each briefly. foundational cloud design vocabulary.
Savings Plans vs Reserved Instances for mixed compute
Compute Savings Plans cover EC2, Fargate, and Lambda flexibly; EC2 Instance Plans and RIs trade flexibility for slightly deeper discounts. commitment-discount strategy. over-committing or ignoring usage variability.
Grant an EKS pod IAM access to S3
IRSA maps a service account to an IAM role via the cluster OIDC provider, and pods exchange a projected token for short-lived STS credentials. secure workload identity. hardcoding keys or sharing the node profile.
RDS Multi-AZ versus Read Replicas
Multi-AZ is a synchronous standby for failover, replicas are async for read scaling, combine both when needed. availability versus scalability. claiming the standby serves reads or that replicas auto-failover.
AWS Fault Injection Simulator
AWS Fault Injection Simulator is a controlled chaos button: it breaks resources on purpose to prove your failover works before real disasters. Run it before peak traffic to validate auto-healing.
Differences between on-demand, reserved, and spot EC2 instances?
Tests cost-reliability-commitment tradeoffs for ML infrastructure. Good answers map on-demand to experiments, reserved for production training, and spot to fault-tolerant batch jobs. Red flag: spot for real-time serving or skipping reserved capacity analysis.
What is the wrong and right way to manage ML database secrets?
This tests secret management hygiene for ML pipelines. A strong answer rejects hardcoded secrets and env vars, then proposes AWS Secrets Manager with IAM retrieval, TLS, caching, and rotation. A red flag is suggesting .env files, ConfigMaps, or CLI arguments.
How would you architect dynamic training resource provisioning?
Routes jobs to right-sized instances, provisions managed jobs using spot, and auto-tears down on completion. elastic ML infrastructure and cost control. suggesting always-on clusters or serverless GPU training.
Design a system for batch scoring millions of customer records daily
Shard jobs via scheduler; partition storage; right-size CPU/GPU on spot; retry. Decoupling orchestration, storage, and compute for batch inference with cost/SLA in mind. One monolithic VM or real-time APIs for batch.
AWS Inferentia and Annapurna Labs
AWS Inferentia is an AWS chip product line. Annapurna Labs, Amazon's semiconductor division acquired in 2015, builds Nitro, Graviton, and Trainium and ranks among TSMC's top five fabless customers. Do not assume Annapurna designs every AWS accelerator.
Design a clickstream pipeline from ingestion to data warehouse
Tests data pipeline design under load: buffering, idempotent transform, and warehouse modeling. A strong answer orders ingestion via Kinesis or MSK, Spark EMR sessionization, and Redshift star schemas. Red flag: no buffer and direct warehouse writes.
AWS Cost and Usage Report (CUR)
Think of CUR as your AWS itemized receipt, delivered daily to S3. It breaks down charges by hour, product, resource, and tag for spreadsheets or Athena queries. Mid-month numbers are estimates, so do not lock budgets until the report finalizes after invoicing.
AWS SDK: Code That Operates Your Cloud
The AWS SDK turns AWS API calls into native code so your app can command S3 or DynamoDB directly. It handles auth, retries, and formatting automatically. The footgun: forgetting region or credentials causes silent failures that look like network errors.
Outline an automated pipeline to load a daily CSV into a database
Event trigger on drop, schema validation, idempotent load, and observability. designing a minimal production-ready ingestion pipeline, not just listing tools.
Enforce GDPR's Right to be Forgotten Across a Complex Architecture
This tests your design of a verifiable, async deletion workflow. A strong answer proposes a central index metastore, an orchestrated workflow (e.g., Step Functions) for deletion, and an auditing layer.
Design a Real-Time Anomaly Detection System for E-commerce Events
This tests your ability to design a real-time data pipeline and apply ML to a business problem. Outline a streaming architecture (e.g., Kinesis), processing, and storage.
Build a pipeline to load a daily CSV into a database
This tests your ability to connect basic cloud services (storage, compute, database) into a simple, event-driven data pipeline. A good answer mentions an event trigger (S3), a serverless function (Lambda), and a database (RDS), plus error handling.
Design a GDPR 'Right to be Forgotten' System
Tests your ability to design a verifiable data deletion system. A good answer outlines an index metastore, an orchestrated workflow, and auditing. A red flag is focusing only on the primary database, ignoring the data lake, warehouse, and caches.
Build a pipeline to load CSVs into a database
Tests your grasp of event-driven architecture and basic ETL. A good answer outlines a trigger (storage event), a processing function (serverless), and a destination (database), mentioning error handling. A red flag is describing a manual or cron-based process.
Shadow Deployment: Test in Production, Safely
Shadow deployment copies live user traffic to a new "shadow" service for testing without user impact. It's used to validate new code versions with real-world load or to analyze traffic for security threats.
Reserved Instances vs. Savings Plans: Pre-pay for Cloud Discounts
Think of Reserved Instances and Savings Plans as buying cloud compute in bulk for a discount. You commit to a certain usage level for 1-3 years to save money on steady-state workloads like production databases.
Amazon SageMaker Pipelines: Repeatable ML Workflows
Think of SageMaker Pipelines as a CI/CD pipeline for ML models, automating workflows from data prep to deployment. Use it for reproducible training and automated retraining.
Slash ML Training Costs with Spot Instances
Spot Instances are like flying standby for compute: you get a massive discount but can be 'bumped' if someone pays full price. Use them for ML training jobs that can be paused and resumed. The footgun is using them for tasks that can't tolerate interruption.
Amazon Bedrock: One API for Many AI Models
Amazon Bedrock is an API gateway for foundation models, letting you switch AI providers without rewriting code. It's used to build generative AI apps while avoiding vendor lock-in.
Get AWS bites daily.
Five a day, five minutes, offline. With quizzes so it sticks.
Open testing — you’ll join as an early tester.