Introduction
When I started Week 2 of the DevOps Micro Internship, I expected to learn a few new AI features and commands. Instead, I discovered a different way of thinking about AI in general terms. This week showed me that AI is not just a tool for answering questions, fine tuning resume’s, or graphic elements. I discovered it can become an active participant in a structured development workflow when guided by the right rules, permissions, and context. Through hands on exercises with Claude Code, Skills, and Memory, I gained a deeper understanding of how Agentic AI fits into modern DevOps practices. It also took my imagination to other areas of learning where it can be productively applied. Along the way, I also learned a lot about my own learning habits, the mistakes I tend to repeat, and the systems I need to build to become a more disciplined engineer.
Biggest Technical Insight I Got
This Week The most significant technical insight I gained this week was understanding how Agentic AI can execute structured workflows instead of simply responding to prompts. Before this week, I tried it out as a coding assistant (even though I am not completely into software development) that generated snippets or answered technical questions. Working with Claude Code changed that perspective completely. One feature that stood out was Skills. I learned how they transform long, repetitive prompts into reusable slash commands that automate common tasks. Running /scaffold-terraform and watching Claude generate an entire Terraform project showed me how AI can streamline infrastructure creation while following predefined instructions. Another important lesson was the value of tool restrictions. The tf-plan skill only has access to the tools it actually needs, reflecting the DevOps principle of least privilege. That small design choice highlighted an important reality: automation is most effective when it is carefully controlled rather than given unrestricted access.Biggest Insight I Got About Myself
This Week This week also taught me something important about how I learn. I realized that I understand technical concepts much better through hands on practice than by reading documentation or watching tutorials alone. Running commands, exploring generated files, and experimenting with Claude Code helped the concepts become much clearer. I also noticed how much the smallest details matter. Carefully following instructions, checking folder structures, and reading command output often determined whether I completed a task successfully or spent unnecessary time troubleshooting. It reminded me that becoming a better engineer isn't just about writing code, it's also about developing patience, attention to detail, and consistency.My Biggest Weakness or Loop I Noticed
One thing I have noticed about myself is that I tend to rush through assignments. I want to get ahead of myself and finish on time, instead ensuring I don’t miss necessary details. This habit sometimes causes problems. For example I forget to include screenshots or check if files are in the right places. Another weakness is how I react to errors. My initial reaction was to assume that something was broken or that I missed a step in the documentation steps instead of carefully reading the error message. During the Terraform exercises, I learned that some failures, such as the AWS authentication errors were expected to occur since there were no credentials to enable access. Instead of indicating failure, they explained exactly what configuration would be needed in a real environment. That experience reminded me that error messages are often valuable learning points rather than obstacles.One System I Will Implement From This Week
The biggest habit I plan to adopt is creating a checklist before starting every assignment. My checklist will include:
Required tasks Commands to execute
Screenshot requirements
Expected outputs
Before I move on to the step I will check that each item on the checklist is done. I will use the checklist at the start of every session in a project run. The checklist will help me stay organized. I will make fewer mistakes and I might not miss things I need to do. I think using the checklist all the time will make me better at following documentation in the course of handling projects.What I Learned About Agentic AI and DevOps
This week fundamentally changed my understanding of both Agentic AI and DevOps. I learned that AI is far more than a conversational assistant. With the right configuration, it can execute structured workflows, remember project specific decisions, automate repetitive tasks, and operate within clearly defined permissions.
I also gained a stronger appreciation for DevOps itself. Automation alone is not enough, it must be supported by documentation, security, permissions, and human oversight. Features such as Skills, Memory, and tool restrictions demonstrated that reliable automation depends on clear processes and responsible access control rather than unlimited AI capabilities.One of my biggest takeaways is that successful DevOps isn't about replacing engineers with AI. It's about combining human judgment with intelligent automation to produce faster, safer, and more consistent results. One of the important things I learned is that DevOps is not about using Agentic AI to replace the people who build and fix things. It is about using Agentic AI as a team where humans act as guardrails to help steer and direct each project to its predefined direction. People work faster and safer when you combine what people are good at with what Agentic AI's good at, you can get really good results.
My Week 2 Highlight
Without question, my favorite moment this week was running /scaffold-terraform and watching Claude generate a complete Terraform project with multiple configuration files. Seeing an entire infrastructure scaffold created from a single command transformed Agentic AI from an abstract concept into something practical that I could immediately apply. Another memorable experience was testing Memory. After closing Claude Code completely and starting a brand new session, Claude correctly remembered the project's hero section colors, mobile breakpoints, and the instruction never to recommend JavaScript for the project. Seeing those decisions persist across sessions demonstrated how valuable project memory can be for maintaining consistency across a development workflow.
Final Thoughts Looking back at the week I realised that I have gained a lot more than technical knowledge. It actually changed how I think about creating things with AI, especially infrastructure. I learned that effective DevOps is built on repeatable systems, clear documentation, secure automation, and continuous improvement. Just as importantly, I also became more aware of how I learn and where I need to be more focused.
As I move on to the next part of this internship I want to learn not just new tools but also how to be a better engineer. Every task, every problem and every success is helping me become a more careful and skilled DevOps engineer.
Week 2 reminded me of something, the best engineers do not just automate tasks, they build systems that work well, learn from every issue and keep improving their work and themselves. As a DevOps engineer I want to keep improving my DevOps skills. The best engineers build systems and continuously improve both their workflows and themselves.
P.S. This post is part of the DevOps Micro Internship (DMI) with Agentic AI — Cohort 3 — by Pravin Mishra. My graded progress is public: https://dmi.pravinmishra.com/s/wisegeorge1.html ·
Start your DevOps journey: https://dmi.pravinmishra.com/?utm\_source=student&utm\_medium=ps-blog&utm\_campaign=cohort3
#DMIByPravinMishra #AgenticAI #ClaudeCode #DevOps #LearningInPublic
