Black Friday SaleIT 360 Bundle complimentary with every master programSee the offer

Cloud

AWS Security Specialty Interview Questions and Answers

SCS-C02 has 65 questions in 170 minutes. Get AWS Security Specialty interview questions with IAM, KMS and incident response answers, plus CLI steps.

By Network Kings · Published 16 November 2026

AWS Security Specialty Interview Questions and Answers

Expect scenario questions, not definitions: IAM evaluation order, KMS envelope encryption, cross-account S3 with a customer managed key, and isolating a compromised EC2 instance. Each model answer below gives the short spoken version first, then the policy JSON or CLI command you would write on a whiteboard. The SCS-C02 exam itself has 65 questions in 170 minutes.

Key takeaways

  • Interviewers probe why a control exists, so answer with the failure mode first, then the control, then the trade-off.
  • An explicit deny always wins, and a permissions boundary or SCP only caps access, it never grants it.
  • Cross-account S3 with a customer managed key needs the bucket policy, the key policy and the caller's IAM policy to all agree.
  • During an incident, preserve evidence first: quarantine, snapshot, capture logs, then clean up.
  • Practise breaking and fixing policies in a real account so you have stories to defend.

AWS for Network Security [ Firewalls ]

In this article

What does the AWS Security Specialty interview actually test?

It tests whether you can reason about failure, not recite service names. The exam behind the credential is AWS Certified Security - Specialty, code SCS-C02, which costs 300 USD and has 65 multiple-choice or multiple-response questions in 170 minutes. You are charged in whatever currency your card uses. You book through AWS Certification and sit it at a test centre or online proctored. If you fail, you must wait 14 calendar days before retaking.

⟦Figure block 1 – kept as is; edit it in the HTML view⟧

FactValue
Exam codeSCS-C02
Fee300 USD
Questions65
Duration170 minutes
Retake wait14 calendar days

A typical loop has a recruiter screen, a technical deep dive with a security architect, a hands-on or design round, and a behavioural round. The five domains are incident response, logging and monitoring, infrastructure security, identity and access management, and data protection. The certificate gets you shortlisted. The lab stories close the round.

For the US market, the US Bureau of Labor Statistics reports a median annual wage of 129,180 USD for information security analysts in May 2025, with 21 percent projected growth from 2025 to 2035. For the UK, UAE, Saudi Arabia, Canada, Australia, Qatar, Kuwait and India, read AWS cloud security engineer salary and skills for regional context.

How does IAM evaluation work, and how do SCPs differ from permission boundaries?

Short version: an explicit deny anywhere wins. Otherwise, the request needs an allow that survives every cap above it, and anything not allowed is an implicit deny. The order to say out loud: explicit deny, then SCP check at the organization level, then permissions boundary, then identity-based or resource-based allow, then default deny.

⟦Figure block 2 – kept as is; edit it in the HTML view⟧

The classic follow-up: a permissions boundary never grants anything. It only sets the maximum an IAM principal can ever get.

Policy typeAttaches toCan grant?
Identity-basedUser, group or roleYes
Resource-basedResource such as a bucket or keyYes
SCPOrganization root, OU or accountNo, caps only
Permissions boundaryOne IAM user or roleNo, caps only
Session policyAn assumed-role sessionNo, caps only

An SCP applies to accounts inside AWS Organizations and touches every principal in them, including the account root user. A boundary applies to individual principals. The gotcha: a deny at the root flows down to every OU and account, while the management account is not restricted by SCPs.

A condition-key example interviewers like: allow only when aws:PrincipalOrgID matches your organization, and deny when aws:PrincipalArn is outside the approved role list.

Common student mistakes are expecting an SCP to grant access and forgetting the deny side is evaluated first. For governance, mention AWS Organizations, Control Tower guardrails, and IAM Identity Center permission sets instead of long-lived IAM users.

How do you prevent the confused deputy problem and set up cross-account S3 access with a CMK?

The confused deputy problem is a third-party service using its own permissions to act on your resource while being tricked into acting for the wrong customer. Fix it with sts:ExternalId in the role trust policy, plus aws:PrincipalArn and aws:PrincipalOrgID conditions where they fit.

{
  "Effect": "Allow",
  "Principal": {"AWS": "arn:aws:iam::VENDOR_ACCOUNT:root"},
  "Action": "sts:AssumeRole",
  "Condition": {"StringEquals": {"sts:ExternalId": "unique-per-customer-id"}}
}

Without the ExternalId, any customer of that vendor who guesses your role ARN can make the vendor assume it.

Cross-account S3 with a customer managed key needs three things: a bucket policy allowing the role, a key policy allowing the role, and an IAM policy on the caller. Miss the key policy and you get AccessDenied even though S3 looks correct.

Debug in this order and read the error code before changing anything:

  1. aws s3api get-bucket-policy --bucket NAME
  2. aws kms get-key-policy --key-id KEY --policy-name default
  3. aws kms describe-key --key-id KEY

Also mention S3 Block Public Access at account and bucket level, and that an explicit deny in a bucket policy overrides any allow. Interviewers usually break one of the three statements and watch how you narrow it down.

How does KMS envelope encryption work, and what is a DEK?

KMS encrypts a data key, the data key encrypts your data, and KMS never sees your payload. That data key is the DEK (data encryption key). You get it in two copies: plaintext for encrypting now, and encrypted for storing beside the ciphertext. You cache the plaintext DEK briefly rather than call KMS for every object, which cuts latency and request volume.

Key typePolicy controlRotationCross-account
AWS ownedNone, AWS manages itAWS handles itNot applicable
AWS managedFixed policy, view onlyAWS handles itNot shareable
Customer managedFull key policy controlAutomatic rotation availableYes, via key policy

Customer managed keys carry per-key and per-request charges, while AWS owned keys add no key charge to you. Rotation has a limit candidates forget: it does not re-encrypt existing ciphertext, and the old backing key stays usable for decrypting.

On S3, SSE-S3 is the better choice when you need encryption at rest without key policy overhead. SSE-KMS adds audit and policy control. SSE-C means you supply the key on every request. A common mistake is giving an application kms:Decrypt on a broad key without condition keys, which quietly defeats the encryption boundary.

How do you isolate a compromised EC2 instance and handle a leaked access key?

Quarantine first, but keep the instance running so memory survives. Swap its security groups to one with no ingress and no egress:

aws ec2 modify-instance-attribute --instance-id i-xxxx --groups sg-quarantine
aws ec2 describe-instances --instance-ids i-xxxx

Do not stop or terminate before capture. Snapshot the EBS volumes into a separate forensic account, then preserve CloudTrail and VPC Flow Logs. Revoke the instance profile's role sessions and check IMDS exposure before you trust anything on that host. Interviewers listen for order of operations: evidence first, cleanup after.

For a leaked access key, deactivate first, investigate, then delete and rotate:

aws iam update-access-key --access-key-id AKIA... --status Inactive --user-name USER
aws cloudtrail lookup-events --lookup-attributes AttributeKey=AccessKeyId,AttributeValue=AKIA...

Then rotate anything that key could reach. Prevent a repeat by moving to IAM roles and IAM Identity Center, and add an SCP that denies iam:CreateAccessKey in production accounts.

Automation helps: an EventBridge rule can trigger a Lambda function to quarantine and notify. Test that runbook in the AWS hands-on labs before a real incident.

What is the difference between security groups, NACLs, Network Firewall and WAF?

Security groups and NACLs filter by IP and port. Network Firewall and WAF inspect deeper. The mistake interviewers catch most: NACLs are stateless, so return traffic needs its own rule, including the ephemeral port range.

ControlStateRulesLevelBest use
Security groupStatefulAllow onlyInstance or ENIDefault workload filtering
NACLStatelessAllow and deny, orderedSubnetSubnet-wide blocks
Network FirewallStateful inspectionDomain and IDS/IPS style rulesVPC trafficRules security groups cannot express
WAFHTTP layerManaged rule groups, rate-based rulesCloudFront, ALB, API GatewayWeb attacks and abuse

Add Route 53 Resolver DNS Firewall for outbound domain control and AWS Shield Advanced for DDoS, and keep that part short.

Your debugging sequence: VPC Flow Logs first, where a missing ephemeral port rule shows up as REJECT on the return path, then NACL rule order, then security group references, then the route table. If VPC fundamentals feel shaky, read what is AWS VPC first.

How do GuardDuty, Security Hub and AWS Config fit together in a SOC?

The one-liner that lands: CloudTrail is the record, GuardDuty is the alarm, Security Hub is the dashboard, Config is the drift detector.

ServiceQuestion it answers
CloudTrailWho called which API, and when?
CloudWatchWhat do metrics and logs show right now?
GuardDutyIs something malicious happening?
Security HubWhat are all my findings and standards failures?
AWS ConfigDid a resource drift from the required configuration?
MacieWhich S3 objects hold sensitive data?
InspectorWhich workloads have vulnerabilities?

Walk a triage flow: a GuardDuty finding lands in Security Hub, an EventBridge rule matches it, a Lambda function acts, and a ticket is created carrying the finding ID. A sample action is removing a public S3 bucket policy or blocking a suspicious key on alert.

Protect the logs too: enable CloudTrail log file validation, use an organization trail, and put S3 Object Lock on the log bucket.

How should you prepare for the interview and the SCS-C02 exam?

Build three stories you can defend: an IAM cleanup, a KMS or S3 data protection fix, and one incident response drill. Practise in a real account. Break a bucket policy, break a key policy, then fix both without reading a blog.

In the last week, revise evaluation logic, envelope encryption, cross-account patterns, and the detection stack. Answer every question: failure mode first, then control, then trade-off. Cloud security engineer and information security analyst roles reward evidence you can explain, not certificate counts.

Next steps: the AWS security course is the pillar page, and the live AWS Cloud Security Master Program adds mentor-led labs. Use the AWS hands-on labs for drills, compare tracks in Azure vs AWS security certification, and check upcoming live batches and fees for dates.

Cloud Security Engineer || AZ-700 & Az500 || Network Kings

Frequently asked questions

How many questions are in the AWS Security Specialty exam?

The SCS-C02 exam has 65 questions, a mix of multiple choice and multiple response, with 170 minutes to finish. The fee is 300 USD per the official AWS certification page.

What is the IAM evaluation order when a deny appears in an SCP and an allow appears in an identity policy?

The deny wins. An explicit deny anywhere ends evaluation, so the identity policy allow never takes effect. Without a deny, the SCP and any boundary must also allow the action, otherwise it falls to implicit deny.

Do I need to know CLI commands for an AWS security engineer interview?

Yes, at least the core ones. Expect to write or read commands for security groups, KMS key policies, CloudTrail lookups and IAM key deactivation. Interviewers want to see you investigate before you change anything.

How do SCPs differ from permissions boundaries in practice?

An SCP caps every principal in the accounts it covers, including the account root, and is managed through AWS Organizations. A permissions boundary caps one IAM user or role. Neither grants permissions, they only limit the maximum.

How do you give a third-party vendor access to your AWS account safely?

Create a role the vendor assumes, with a trust policy that requires a unique sts:ExternalId. Add aws:PrincipalArn or aws:PrincipalOrgID conditions where they fit, and grant only the permissions the vendor needs.

How long should I prepare before booking the SCS-C02 exam?

It depends on your hands-on AWS experience, and no official figure exists. Prepare until you can break and fix IAM, KMS and cross-account S3 policies unaided, then book. If you fail, the retake wait is 14 calendar days.

Learn this live in the AWS Cloud Security Master Program

Live classes with engineers from top companies, 24x7 labs and a counsellor to plan your path.

Why learn from the top 1% of engineers?

Because passing an exam and doing the job are different skills, and only experience closes the gap. Network Kings mentors have spent 11 to 20+ years building and running networks, cloud platforms and pipelines at companies like Apple, Juniper Networks, IBM and CDAC. They teach what breaks in production, why, and how engineers fix it under pressure.

  • Atul Sharma, Founder & CEO mentor at Network KingsAtul Sharma, Founder & CEO
  • Lazaro (Laz) Diaz, Networking mentor at Network KingsLazaro (Laz) Diaz, Networking
  • Sanjay Sharma, AWS mentor at Network KingsSanjay Sharma, AWS
  • Vikas Kumar, Azure mentor at Network KingsVikas Kumar, Azure
  • Ajay, Azure mentor at Network KingsAjay, Azure
  • Amit Bhatt, Linux & DevOps mentor at Network KingsAmit Bhatt, Linux & DevOps

Atul Sharma (Founder & CEO, 12+ yrs), Lazaro (Networking), Sanjay Sharma (AWS, 20+ yrs) and more. Meet the mentors

Same problem, two answers. Pick a track:

On the job: Two routers' OSPF neighbours are stuck in ExStart and never reach Full.

The textbook answer

OSPF moves through Down, Init, 2-Way, ExStart, Exchange, Loading and Full. Memorise the neighbour states for the exam.

Correct. Not enough to fix it.

How a production engineer teaches it

  1. Suspect an MTU mismatch first. ExStart and Exchange stalls usually come down to it.
  2. Compare show ip ospf interface on both ends and fix the interface MTU.
  3. Use ip ospf mtu-ignore only as a stopgap, and say so in the change record.
  4. Write down why it happened, so the next engineer finds it in minutes.
  • The why behind every command, not just the syntax.
  • Hands-on labs for every topic, so the first fault you fix is not at work.
  • Doubts answered by someone who has fixed it in a live network, cloud or pipeline.