What the roles are
One family of roles, and a door in from each background
Cloud and DevOps is a family of roles that provision infrastructure as code and ship software through automated pipelines. In a 4,304-posting sample the titles split DevOps Engineer 31%, Site Reliability Engineer 15%, Platform Engineer 12%, Infrastructure Engineer 9%. Each has a natural starting point in the operations world.
Cloud engineer
Provisions and runs cloud environments (VPC, compute, storage, IAM) with Terraform; migrations; cost. Cloud network engineers do the VPC, routing, transit and load-balancer design, all as code.
DevOps engineer
CI/CD pipelines, containers and Kubernetes, configuration management, release automation. The most common title, 31% of postings, and the one Infosys, Capgemini and TCS post most in India.
SRE and database reliability
Availability and latency targets (SLOs), monitoring and alerting, incident response, on-call. For DBAs: managed databases (RDS, Aurora, Cloud SQL), HA/DR and automation with Terraform and Ansible.
What transfers, by background
- Sysadmin and IT ops: Linux, Bash, networking basics, troubleshooting, on-call. Monitoring is named in 39% of both sysadmin and cloud engineer postings; disaster recovery in 13% of both.
- Network engineer: subnets become VPC subnets, routing becomes route tables, firewalls become security groups, segmentation becomes NACLs and WAFs, access control becomes IAM. Load balancers and DNS carry over directly.
- DBA: backup and recovery, HA and failover, performance tuning, security and auditing are the first things Rackspace and Adobe list. The additions are managed services, Terraform and Ansible, containers and SRE practice.
- What does not transfer at par: Windows Server, Active Directory and desk support are the sysadmin skills with the lowest pay effect in the same study; Azure DevOps, Windows and PowerShell sit in the lowest DevOps pay tier.
InterviewStack, Cloud Engineer Academy, Rackspace and Adobe postings.
"Provision and manage AWS cloud infrastructure with Terraform. Automate infrastructure deployment and configuration with Ansible or Chef. Manage containerized applications with Docker and Kubernetes."
Infosys, DevOps Engineer, Bengaluru, posted 27 Oct 2026Who it is for
Built for people who run production and are ready to automate it
Entry-level is 2–8% of postings in every study we read; the common bar is three to five years, and India postings sit at three to ten. Operations experience is what the bar is made of. The postings want it paired with automation.
✓A fit if you
- Administer Linux or Windows servers and already script in Bash or PowerShell
- Work in IT operations or a NOC with real on-call and incident experience, which SRE loops test directly
- Are a network engineer who knows routing, firewalls, DNS and load balancing
- Are a DBA with backup, recovery, HA and tuning experience on PostgreSQL, MySQL, Oracle or SQL Server
- Have automated at least one task you used to do by hand, and want to do that for everything
- Can explain an incident end to end: detection, mitigation, root cause, what changed afterwards
…Not yet if you
- Are a fresher with no production exposure: entry roles are 2–8% of postings. Start with a support, NOC or junior admin role
- Work tickets only and have never automated anything: every posting lists Terraform or Ansible plus scripting as required
- Are counting on a certification alone: Infosys lists none; Adobe and Rackspace list them as preferred, beside hands-on IaC and Kubernetes as required
- Have only console-clicked a cloud account, with no code behind it
Transition guides put a working sysadmin at roughly 60% of a DevOps skill set before starting. Treat that as a vendor claim, but the posting data agrees on where the other 40% is. DevOpsBoys
Experience bars in current India postings
| Employer, role | Location | Years required | Also asks for |
|---|---|---|---|
| Infosys, DevOps Engineer | Bengaluru | 3–5 | Kubernetes, Docker, AWS (EC2, VPC, IAM, S3, RDS, CloudWatch), Terraform, Ansible or Chef, Jenkins, Linux and shell; preferred EKS, Python, DevSecOps, Prometheus/Grafana |
| Roche, DevOps DBA Engineer | Pune | 4+ | AWS (EC2, EKS, S3, VPC, Route 53, RDS); one of RDS, Oracle, MySQL, MSSQL, Redshift; Terraform and Ansible; ELK, CloudWatch; patching, releases |
| Adobe, Database Reliability Engineer | Noida | 4–7 | PostgreSQL on RDS/Aurora in production; HA/DR, zero-downtime upgrades; Terraform, Ansible, Docker/Kubernetes in CI/CD; Prometheus, Grafana; on-call |
| IT services firm, Senior AWS DevOps Engineer | Hyderabad | 5–8 | AWS across compute, networking, serverless, containers; Jenkins or GitHub Actions; Terraform, CloudFormation, Ansible; EKS, Helm; on-call to a 99.99% SLA |
| Rackspace, Cloud DBA III (PostgreSQL) | Hyderabad | 6–10+ | Two of AWS, Azure, GCP; Terraform, CloudFormation, Kubernetes; SLOs, incident management, 24x7 rotation |
Postings read between 30 Oct and 5 Nov 2026 via company sites and job boards. The Roche posting and the Hyderabad senior role are closed and quoted for their requirements, which recur across similar roles.
Skills to learn
Five skills make up the gap
Containers and Kubernetes, CI/CD, infrastructure as code, one cloud platform, and scripting. Everything else the postings ask for, Linux, networking, monitoring, databases, you already have. The two charts show the same skills on each side of the move.
Share of sysadmin postings naming each skill
4,263 systems administrator postings, July 2026. InterviewStack
The same skills in cloud engineer postings
3,766 cloud engineer postings, same study. In 7,523 DevOps postings the top asks are CI/CD 67%, Kubernetes 56%, AWS 54%, Python 53%, Terraform 50%. InterviewStack
Linux, scripting and one cloud
Deepen what you have, add one platform
- Linux administration and shell; Python or Bash for automation, Git for everything
- One cloud, properly: AWS is named most (EC2, VPC, IAM, S3, RDS, CloudWatch, EKS, Route 53). In India postings Azure is mentioned in 19%, AWS 17%, GCP 9%, so Azure is a valid first cloud for Microsoft-shop admins
- Networking in the cloud: VPC design, security groups, load balancers, DNS, private connectivity
- IAM least privilege, secrets, encryption: security appears in 57% of DevOps postings
Infrastructure as code and configuration management
Write the environment down
- Terraform: modules, remote state with locking, separate dev and prod, plan in CI before merge
- CloudFormation or Bicep where the shop is single-cloud
- Ansible or Chef for configuration: both named in the Infosys, Roche and Adobe postings
- The habit behind the tools: no change by hand that is not also in code
Containers, Kubernetes and CI/CD
Ship through a pipeline
- Docker first; then Kubernetes administration and deployment, EKS or AKS, Helm
- Jenkins, GitHub Actions or GitLab CI; ArgoCD for GitOps
- Blue-green and canary releases, rollbacks, zero-downtime deployments
- Docker and Kubernetes co-occur in 33% of DevOps postings; CI/CD and Kubernetes in 46%
Observability, reliability and cost
Run it to a target
- Prometheus, Grafana, ELK, CloudWatch, Datadog; alerts that fire on what matters
- SLOs and SLIs, incident response, on-call, runbooks, postmortems: the highest-paying tier in the DevOps study
- Cost: right-sizing, reserved capacity, tagging; one senior posting lists "automate provisioning, scaling and cost optimisation"
- For DBAs: RDS/Aurora, Cloud SQL, Azure SQL; HA/DR drills; zero-downtime upgrades; databases in containers
Skill groups compiled from the Infosys, Roche, Adobe, Rackspace and Hyderabad postings and the InterviewStack, JobGlance and PrepNPlaced posting analyses.
Projects to build
Rebuild what you already run by hand, as code
The projects that match these postings take something you operate today and put it in Terraform, behind a pipeline, with monitoring and a documented failure. Hiring managers read the README and the failure story, and skip tutorial deployments. Pick the one that matches your background and target role.
Sysadmin or network → cloud engineer
A full cloud environment as code
A VPC with public and private subnets and NAT, compute with auto scaling, a private managed database, a load balancer, HTTPS via DNS and certificates.
Must include: Terraform modules; remote state with locking; separate dev and prod; least-privilege IAM; terraform plan in CI before merge; an architecture diagram. Network engineers: add transit or peering between two VPCs and private connectivity to a managed service.
Sysadmin or IT ops → DevOps engineer
End-to-end CI/CD with GitOps
Build, test and push an image with GitHub Actions; deploy with ArgoCD to a Kubernetes cluster (kind, k3s or a free tier).
Must include: separate app and GitOps repositories; the documented flow from pull request to sync; image scanning; a rollback you have actually run; liveness and readiness probes.
IT ops, NOC or DBA → SRE and database reliability
An observability stack and a self-healing service
Prometheus, Grafana and logs around a service that can fail in several ways; alerts that fire during a simulated failure; a runbook per alert.
Must include: alert routing; HPA and probes; a screenshot of an incident detected and fixed. DBAs: swap in RDS or Aurora PostgreSQL provisioned with Terraform, a tested failover, point-in-time recovery, and a restore drill with timings.
The three artefacts to walk into an interview with
- A public repository with a README: architecture diagram, the problem, stack choices and why, working run commands, and what broke.
- A write-up of one failure and how you debugged it. The sources call this the most convincing single item, and it is the thing a sysadmin can write better than anyone.
- A short demo video or live endpoint linked from your resume, so a reviewer can see it without cloning anything.
Sequencing from the transition guides: containers and Git first, then a pipeline, then one cloud, then Terraform, then Kubernetes depth, then the portfolio. About six months part-time for a working admin.
What does not count
- "Deploy a website to S3" and other unmodified tutorials
- Console-clicked environments with no code behind them
- Projects without a README, or that you cannot explain in an interview
- Certification badges with no repository beside them
- A cluster you cannot say how you would upgrade, back up or recover
Match the project to the role: Terraform for cloud engineer, CI/CD for DevOps engineer, observability and self-healing for SRE, the database platform for DBRE.
The interview loop
Scenarios, not puzzles
DevOps and SRE loops run four to six weeks and are built around things that happen in production: an incident walkthrough, a Kubernetes or Linux debugging exercise, a pipeline or infrastructure design, and an on-call conversation. Your operations years are an advantage in every round except the design one, which the projects above prepare you for.
Recruiter screen
Compensation, on-call expectations, tooling. Recruiters filter on production Kubernetes experience, so your project cluster has to be real.
Hiring-manager screen
"Walk me through the worst production incident you have led the response on." Detection, mitigation, resolution, postmortem, with numbers.
Technical screen: scripting, Linux, containers
"Write a script that reports the top five errors in the last 1,000 log lines." "A pod is stuck in CrashLoopBackOff: how do you debug it?" "A service responds by IP but not by hostname: where do you look first?"
Technical deep dive or take-home
CI/CD design, Kubernetes troubleshooting, a Terraform module review, fixing a Dockerfile, writing an Ansible playbook. "Several engineers run Terraform against the same environment: how do you manage state?"
Infrastructure design
"Design a CI/CD platform for 10 services and 50 engineers." "A Terraform module structure for dev, staging and prod." "Reduce the AWS bill by 30% without hurting performance." "A multi-tenant Kubernetes cluster."
Reliability and behavioural
"What do you monitor, and what do you alert on?" "p99 latency is rising while p50 is normal." "A time you disagreed with a developer about a production issue." "A team wants to ship without observability." On-call philosophy.
What strong answers do
- Structure every scenario as diagnose, isolate, fix, prevent recurrence
- Bring one real incident with numbers: time to detect, time to mitigate, what changed afterwards
- Explain trade-offs in design: blue-green versus canary, managed versus self-hosted, cost versus availability
- Treat security and cost as part of every answer, not a separate round
How to prepare
- Have two incident stories ready, one you led
- Practise Linux and Kubernetes debugging out loud on a real cluster
- Be able to read and critique a Terraform module and a pipeline definition
- Know your own project's failure story cold
Rounds and questions from CV-by-JD and Mocklingo.