I've worked in IT for about 15 years, almost entirely on-premises. After handing over my previous BAU responsibilities, I was moved into an Azure DevOps security role with very little preparation or support. I'm now expected to work with Git Bash, pipelines, infrastructure as code, Terraform, and Jira, all of which are new to me.
I'm struggling to connect training videos with real tickets. For example, I picked up a VM decommissioning task and was told to remove a lock by changing the configuration in Azure DevOps and then committing the change through a pipeline. I don't yet know which repository or pipeline is involved, and I haven't even been provided with Git Bash or an IDE. What would be a sensible way to approach this transition without making a dangerous change or falling behind?
5 Answers
A ticketing system is mostly about visibility: assign the ticket, move it to in progress, write down what you checked, and record blockers. If a task involves deleting or changing infrastructure and the request is unclear, mark it as blocked or needing clarification rather than improvising. A good team should prefer a delayed, reviewed change over an unreviewed production incident.
Cloud platforms can make familiar infrastructure feel much more complicated because the controls are spread across templates, YAML, identities, state files, and approval gates. Try translating each cloud object back into something you already understand: a VM, a network rule, a permission, or a deployment step. That usually makes the terminology less intimidating.
For Git Bash, concentrate on the small set of commands your team uses daily—clone, status, pull, checkout, add, commit, and push. You don’t need to learn every command before contributing.
Your 15 years of on-premises experience are not wasted. Networking, operating systems, virtualization, access control, change management, and troubleshooting still apply in Azure. The new part is that those concepts are represented through code and automated workflows.
A practical learning order would be: basic Git operations, the team’s existing Terraform layout, one Azure DevOps pipeline from start to finish, and then the Jira workflow. Read real files that manage infrastructure similar to yours instead of trying to master every generic tutorial first. Keep a small glossary of the team’s terms and ask for a short pairing session on one real ticket.
Don’t try to learn Azure, Terraform, pipelines, Git, Jira, and security all at once. Start with the exact workflow your team uses for one small change. Ask someone to show you which repository contains the Terraform, which branch and pull-request process they expect, which pipeline runs it, and whether the change requires a plan, review, or approval.
For the VM task, don’t simply comment out a lock or edit a pipeline blindly. First understand where the lock is defined, whether it is protecting production, and what the expected change and rollback are. Once you have the right tools and repository access, make a small branch, run the relevant validation or Terraform plan, and have somebody review it before applying anything. Your infrastructure experience is valuable; the main gap is learning how this team expresses and deploys it as code.
Microsoft Learn is a good structured resource for Azure and Azure DevOps. For infrastructure as code, focus on the tool your team actually uses. Since they use Terraform, learn its basic workflow: inspect the configuration, initialize safely, create a plan, review the plan, and only then apply through the approved pipeline. Don’t spend time learning Pulumi or several alternative tools unless your role specifically requires them.
Also, being asked to work immediately without a computer, Git tools, repository access, or onboarding is an organizational problem. Tell your manager clearly what access and training you need, and ask for a named person to review your first few changes.

That makes sense. I’ve been told to make the change before anyone has shown me the repository or process, so I’ll ask specifically for the repo, pipeline, expected plan output, and approval steps rather than guessing.