Interview Prep — Question Index

This page consolidates every Common Interview Questions block across the site into a single rapid-recall index. Each link below jumps to a page that ends with a <details> Q&A section covering the topic. Use this when prepping for a phone screen, a system design round, or a recruiter call where you need to brush up on a specific service or pattern in minutes rather than hours.

Pages are grouped topically — LLM Engineering and ML foundations first, then the data platforms (Databricks, Snowflake), followed by AWS service families (AI/ML, Compute, Security, Database, Serverless), and finally end-to-end System Design.

1. AI / ML — LLM Engineering

Modern LLM-app patterns: retrieval, agents, evaluation, serving, and the surrounding tooling for production RAG systems.

• Agents and MCP — agent loops, tool use, Model Context Protocol
• Function Calling and Structured Output — JSON schema, tool routing, validation
• Hybrid Search and Reranking — BM25 + dense fusion, cross-encoder rerank
• LangGraph and DSPy — state machines, prompt compilation, optimizers
• RAG Evaluation with RAGAS — faithfulness, context precision, answer relevance
• vLLM and Quantization — PagedAttention, GPTQ, AWQ, throughput tuning
• How Vector Embeddings Work in RAG — embedding models, similarity, chunking strategy
• RAG from Scratch — ingest pipeline, retrieval, prompt assembly
• Weaviate Vector Database — HNSW, hybrid filters, schema design
• MyDocumentIntelligence Architecture — end-to-end RAG product, doc ingest, scaling

2. AI / ML — Foundations

Classical deep-learning building blocks that still come up in screens for ML roles.

• LSTM Networks — gates, vanishing gradients, vs. Transformer
• Neural Network Sentiment Classifier — embeddings, loss, training loop

3. Databricks

Lakehouse architecture, Delta Lake internals, PySpark performance, and Mosaic AI.

• Bronze Layer — raw ingest, Auto Loader, schema evolution
• Silver Layer — cleansing, conformed dimensions, SCD
• Gold Layer — aggregations, BI marts, semantic models
• Delta Live Tables — declarative pipelines, expectations, CDC
• Mosaic AI Vector Search — managed embeddings, sync indexes, hybrid query
• Liquid Clustering — vs. Z-order, partition design, write amplification
• Performance Optimization — AQE, broadcast joins, file sizing, caching
• PySpark Deep Dive — Catalyst, Tungsten, RDD vs. DataFrame, UDFs
• PySpark Data Cleaning — null handling, dedup, type coercion
• Databricks Asset Bundles — CI/CD, env promotion, bundle config
• Unity Catalog Setup — metastore, lineage, fine-grained access control

4. Snowflake

Snowflake-specific architecture, performance, and governance topics. (No standalone Snowflake pages currently carry the Q&A block — see Databricks for cross-platform lakehouse questions.)

5. AWS — AI / ML

Managed AI services for vision, language, and document understanding.

• AWS AI Services Overview — service map, when to use which, pricing models
• Amazon Comprehend — NER, sentiment, custom classifiers, PII redaction
• Amazon Rekognition — image/video labels, face match, content moderation
• Amazon Textract — OCR, forms, tables, async vs sync API

6. AWS — Compute / Storage / Networking

Core infrastructure primitives behind almost every AWS system-design answer.

• Amazon EC2 — instance families, Spot vs On-Demand, ENI limits, EBS
• Auto Scaling EC2 and EMR — target tracking, lifecycle hooks, warm pools
• AWS EMR — cluster types, Spot fleet, EMR on EKS
• ECS, EKS, and Fargate — orchestration trade-offs, task vs pod, networking modes
• Amazon S3 — storage classes, consistency, lifecycle, encryption
• S3 Transfer Acceleration — CloudFront edges, multipart, when worth it
• Amazon VPC — subnets, NAT, peering, Transit Gateway, endpoints
• Amazon CloudFront — cache behaviors, signed URLs, origin failover
• API Gateway — REST vs HTTP vs WebSocket, throttling, usage plans
• Route 53 — routing policies, health checks, failover, alias records

7. AWS — Security / Monitoring

Identity, secrets, audit, and observability primitives that frequently anchor a security-focused interview round.

• IAM — policies, roles, trust, condition keys, least privilege
• KMS — CMK vs data keys, envelope encryption, grants
• Secrets Manager — rotation, vs Parameter Store, cross-account access
• CloudTrail — mgmt vs data events, multi-region trails, integrity
• AWS Config and Inspector — rules, remediation, vuln scanning
• CloudWatch — metrics, logs, alarms, composite alarms
• CloudWatch Events / EventBridge — rules, schedules, cross-account buses
• CloudFormation — stacks, drift, change sets, nested templates

8. AWS — Database / Analytics / Serverless

Managed data stores, ETL/analytics, and event-driven compute — the backbone of most data-engineering rounds.

• Amazon RDS — engines, Multi-AZ vs read replicas, backups
• Amazon Aurora — storage layer, Global Database, Serverless v2
• Amazon DynamoDB — partition keys, GSIs, streams, on-demand vs provisioned
• Amazon ElastiCache — Redis vs Memcached, cluster mode, eviction
• Amazon Athena — partitioning, Iceberg, workgroups, cost control
• Amazon QuickSight — SPICE, RLS, embedded analytics
• AWS Glue Data Catalog — crawlers, schema discovery, Lake Formation
• AWS Glue ETL Service — DynamicFrames, job bookmarks, workers
• AWS Glue Workflow — triggers, blueprints, orchestration
• Kinesis Data Firehose — buffering, dynamic partitioning, transforms
• Lambda — cold starts, concurrency, layers, provisioned
• Step Functions — Standard vs Express, retry/catch, callback patterns
• Kinesis Data Streams — shards, KCL, enhanced fan-out, ordering

9. System Design

End-to-end AI system design questions — the kind that show up in senior staff and architect interviews.

System Design Q&A blocks live on AI Systems pages — see the Design / AI Systems index for the full set of architecture diagrams (Sequence, ERD, DFD) covering the same systems.

10. How to Use This Index