AWS Certification
Path 2026

Conceptos de nube (IaaS/PaaS/SaaS) · Infraestructura global AWS (Regiones, AZs, Edge Locations) · IAM (usuarios, grupos, políticas, MFA) · Compute (EC2, Lambda) · Storage (S3, EBS, EFS) · Bases de datos (RDS, DynamoDB) · Redes (VPC básico, CloudFront, Route 53) · Precios y facturación (Free Tier, Cost Explorer, Budgets) · Shared Responsibility Model · Well-Architected Framework (6 pilares) · AWS Support Plans

Conceptos de IA/ML/DL (supervisado, no supervisado, reinforcement learning) · Generative AI (LLMs, foundation models, RAG, prompt engineering) · Amazon Bedrock (modelos fundacionales, guardrails, knowledge bases, agents) · AWS AI/ML services: Rekognition · Comprehend · Transcribe · Polly · Translate · Textract · Forecast · Personalize · Kendra · SageMaker Canvas · Responsible AI (sesgo, explicabilidad, fairness, privacidad) · Seguridad en IA (toxicidad, alucinaciones, PII)

Well-Architected Framework (Operational Excellence, Security, Reliability, Performance, Cost Optimization, Sustainability) · VPC (subnets públicas/privadas, NACLs, Security Groups, NAT Gateway, VPC Endpoints) · EC2 (tipos, AMIs, Auto Scaling Groups, Placement Groups) · S3 (storage classes, lifecycle policies, versioning, replication, presigned URLs) · RDS / Aurora (Multi-AZ, Read Replicas, backups) · DynamoDB (GSI/LSI, streams, DAX) · Lambda + API Gateway (serverless patterns) · ELB (ALB, NLB, GLB) · CloudFront + Route 53 (routing policies) · IAM (políticas de mínimo privilegio, roles, STS) · CloudWatch + CloudTrail · Patrones de alta disponibilidad y DR (RPO/RTO)

Lambda avanzado (execution context, layers, versiones/aliases, cold start, concurrencia, SnapStart) · API Gateway (REST vs HTTP vs WebSocket APIs, Lambda Proxy integration, authorizers) · DynamoDB desde código (SDK, operaciones CRUD, expresiones, optimistic locking) · SQS (standard vs FIFO, visibility timeout, DLQ, long polling) · SNS (fan-out pattern, message filtering) · Kinesis Data Streams (shards, partition keys, KCL, checkpointing) · Elastic Beanstalk (deployment policies) · ECS (task definitions, services, ECR) · CodeCommit · CodeBuild · CodeDeploy (deployment strategies: blue/green, canary, linear) · CodePipeline · CloudFormation (intrinsic functions, change sets, cfn-init) · X-Ray (tracing, sampling rules, segments/subsegments) · AWS SDK: paginators, waiters, retry con exponential backoff · Cognito (User Pools, Identity Pools, JWT)

CloudWatch (métricas custom, dashboards, Logs Insights, alarmas compuestas, Synthetics) · AWS Systems Manager (Session Manager, Parameter Store, Patch Manager, Automation runbooks, OpsCenter) · AWS Config (rules, conformance packs, remediation automática) · CloudTrail (trails multi-región, integración con EventBridge) · VPC Flow Logs + análisis · Auto Scaling avanzado (target tracking, step scaling, scheduled) · Backup (planes, vault lock) · Cost Explorer + Budgets + Cost Anomaly Detection · Trusted Advisor · AWS Organizations (SCPs básicos) · EventBridge (event bus, rules, targets) · Incident Manager · Estrategias de patch management y compliance

S3 (data lake patterns, particionamiento, storage classes para datos fríos) · AWS Glue (catalogación, ETL jobs PySpark/Python, Glue Studio, Data Quality, triggers) · Athena (SQL sobre S3, Workgroups, partitioned tables, output formats: Parquet/ORC) · Amazon Redshift (distribución de tablas, sort keys, carga masiva COPY, Spectrum, clustering) · Kinesis Data Streams + Kinesis Firehose (procesamiento en tiempo real, transformación Lambda) · Amazon EMR (Spark, Hadoop en clúster managed, cost optimization con Spot) · AWS DMS (migración de bases de datos, CDC) · Lake Formation (permisos a nivel de columna, row-level security, tags) · Step Functions (orquestación de pipelines) · QuickSight (visualización básica) · EventBridge Pipes · Principios de modelado dimensional (star schema, snowflake)

SageMaker core (Training Jobs, Processing Jobs, Feature Store, Model Registry, Endpoints — real-time vs async vs batch) · SageMaker Pipelines (MLOps, pasos de pipeline: procesamiento, entrenamiento, evaluación, registro, despliegue) · Estrategias de despliegue ML (A/B testing, shadow mode, canary, blue/green en SageMaker) · Model Monitor (data drift, model quality, bias drift, explicabilidad con Clarify) · Bedrock Fine-Tuning y Continued Pre-training · Feature engineering a escala (SageMaker Data Wrangler, Glue) · Evaluación de modelos (confusion matrix, AUC-ROC, RMSE, selección de métricas) · Optimización de entrenamiento (hiperparámetros, AMT, SageMaker Experiments) · Seguridad en ML (VPC isolation, KMS, IAM roles por step) · Inferencia eficiente (Inference Recommender, quantización, model serving)

AWS Organizations (multi-account, SCPs, OU design, Control Tower, Account Factory) · Hybrid networking (Direct Connect: VIF types, LAG, resiliency; Site-to-Site VPN, AWS VPN CloudHub) · Advanced networking (Transit Gateway: peering, multicast, route domains; AWS PrivateLink; VPC Sharing) · Disaster Recovery (Backup & Restore, Pilot Light, Warm Standby, Multi-Site Active/Active; RTO/RPO analysis) · Migration strategies (6 Rs: Rehost, Replatform, Repurchase, Refactor, Retire, Retain; Migration Hub; DMS; Application Discovery Service) · Cost optimization avanzado (Savings Plans, Reserved Instances vs On-Demand vs Spot, Compute Optimizer) · Event-driven architecture (EventBridge, Step Functions Express/Standard, SQS + SNS fanout) · Security avanzado (AWS SSO/IAM Identity Center, permission boundaries, SCPs, detective controls con Security Hub + GuardDuty) · Arquitecturas de datos a escala (Lake Formation, Redshift + Spectrum, OpenSearch)

SDLC automatizado (CodeCommit, CodeArtifact, CodeBuild: buildspec.yml, CodeDeploy: AppSpec, deployment hooks, CodePipeline: stages, actions, approvals, manual gates) · IaC avanzado (CloudFormation: nested stacks, macros, custom resources con Lambda, cfn-guard; CDK: constructs L1/L2/L3, aspects, testing) · ECS/EKS deployment strategies (rolling, blue/green via CodeDeploy, canary) · Monitoring y observabilidad (CloudWatch Container Insights, Contributor Insights, Logs Insights ML anomaly detection, Evidently para feature flags) · AWS Config Rules avanzado (remediation automática, conformance packs, multi-account aggregation) · Security Hub (findings, custom insights, auto-remediation) · Systems Manager avanzado (State Manager, Maintenance Windows, Inventory, Documents) · Incident Manager (response plans, escalation, runbooks) · Event-driven automation con EventBridge + Lambda · Chaos Engineering en AWS (Fault Injection Simulator)

Amazon Bedrock avanzado (Model APIs, Converse API, Streaming, Extended Thinking) · Bedrock Agents (action groups, orchestration, code interpreter, memory) · Bedrock Knowledge Bases (vector store con OpenSearch Serverless o Pinecone, chunking strategies, hybrid search, reranking) · Bedrock Guardrails (toxicity, hallucination grounding, PII redaction, topic denial) · RAG arquitectura (indexación, embedding models, retrieval strategies, re-ranking) · Bedrock Model Evaluation (benchmark datasets, custom metrics) · Fine-tuning y Continued Pre-training en SageMaker · Prompt Engineering avanzado (chain-of-thought, few-shot, tool use, structured output) · Seguridad y governance de GenAI (prompt injection, jailbreak prevention, data privacy) · LLM Observabilidad (tracing con X-Ray, logging de prompts, cost monitoring por token)

IAM avanzado (SCPs, permission boundaries, ABAC con tags, session policies, SAML 2.0 federation, Web Identity federation, cross-account assume role) · KMS (CMKs, key policies, grants, envelope encryption, multi-Region keys, CloudHSM) · Secrets Manager vs Parameter Store (rotación automática, VPC endpoint) · WAF (rule groups, managed rules OWASP, rate limiting, geo-blocking) · Shield Advanced (DDoS protection, proactive engagement) · GuardDuty (findings types, threat intelligence, suppression rules, multi-account) · Inspector v2 (EC2, ECR, Lambda scanning) · Security Hub (standards: FSBP, CIS, PCI-DSS, custom insights, findings workflow) · CloudTrail avanzado (Insights, Lake, multi-región, S3 event logging) · VPC Security (NACL vs SG, VPC Flow Logs, Traffic Mirroring, Private DNS) · Macie (PII detection in S3) · Detective (security investigation) · Incident Response en AWS (forensics: snapshot, isolate, investigate; IR playbooks) · Compliance (AWS Artifact, config conformance packs)

VPC avanzado (CIDR design, sharing, inter-VPC connectivity, IPAM) · Transit Gateway (route tables, peering inter-región, multicast, Network Manager) · AWS PrivateLink (endpoint services, NLB, Interface Endpoints) · Direct Connect (Virtual Interfaces: public/private/transit, LAG, MACsec, resiliencia: single/dual location, AWS Direct Connect Gateway) · AWS VPN (Site-to-Site, Client VPN, VPN CloudHub, aceleración con Global Accelerator) · Route 53 avanzado (DNSSEC, Resolver endpoints, Resolver rules, split-view DNS, latency/geo/weighted routing) · ELB profundo (ALB: path/host/header routing, target group weighting, sticky sessions; NLB: cross-AZ, TLS passthrough, UDP; GWLB: inline inspection) · Network performance (Placement Groups, EFA, SR-IOV, jumbo frames, MTU 9001) · IPv6 en AWS (dual-stack, Egress-Only IGW) · Network security (Network Firewall: stateful rules, Suricata IDS; Gateway Load Balancer para inspection appliances)

Preparación de datos ML (feature engineering, imputación, normalización, one-hot encoding, PCA, label encoding; SageMaker Data Wrangler; Glue DataBrew) · Algoritmos ML core (XGBoost, Linear/Logistic Regression, KNN, Random Forest, k-means, LDA, PCA; cuándo usar cada uno) · Redes neuronales profundas (CNN para visión, RNN/LSTM para secuencias, Transfer Learning) · SageMaker Training avanzado (Distributed Training: data parallel + model parallel, Spot Training con checkpoints, SageMaker Debugger, Profiler) · Hyperparameter Optimization (Bayesian, Random, Grid search con AMT) · SageMaker Deployment (real-time, async, serverless, batch transform, multi-model endpoints, shadow testing, A/B) · Model Monitor (data drift, model quality, bias Clarify, explainability SHAP) · SageMaker Pipelines (MLOps completo) · Rekognition · Comprehend · Forecast · Personalize · Transcribe/Polly/Translate (cuándo usar managed services vs custom) · Seguridad en ML (VPC mode, KMS, IAM por step)