DEV Community: Prashant Lakhera

📌 Enterprise week: Multi-AWS-account support for DevOps Open Agent📌

Prashant Lakhera — Mon, 20 Jul 2026 16:43:37 +0000

📌 Enterprise week: Multi-AWS-account support for DevOps Open Agent📌

This week our focus on DevOps Open Agent is enterprise readiness.

In production, almost nobody lives in a single AWS account. You have prod, staging, security, sandbox and when something breaks, the AI has to investigate the right account, not whatever credentials happen to be on the laptop.

That’s why I’m happy to share: DevOps Open Agent now supports multi-AWS-account investigations.

How it works

✅ Keep hub credentials on the host (~/.aws / IAM) — no packing long-lived keys into every job

✅ Configure target accounts under Integrations → AWS Accounts (account ID + role ARN, optional external ID)

✅ The agent uses STS AssumeRole to enter each account for investigate and topology

✅ The account picker shows your hub identity plus every configured account

✅ Wrong/unknown account IDs fail closed, no silent fallback to the wrong account

So you can run the same AI root-cause flow across the accounts you actually operate, the way real teams work.

Open source, self-hostable, built for DevOps engineers who live in multi-account AWS.

If you’re wrestling with one AI, many accounts, I’d love your feedback on what roles and guardrails you’d want next.

🔗 GitHub: https://lnkd.in/giri6MF5

🎥 YouTube: https://lnkd.in/g2PbG3Hc

🎉 August enrollment is now open at IdeaWeaver AI Labs!
We have 3 programs starting this month, pick the one that fits where you are:
🔵 GenAI for DevOps Engineers
👉 https://ideaweaver.ai/#courses/genai-for-devops-engineers
🟢 GenAI for Beginners
👉 https://ideaweaver.ai/#courses/genai-for-beginners
🟡 NVIDIA-Certified Associate: AI Infrastructure and Operations
👉 https://ideaweaver.ai/#courses/nvidia-certification

🚀 Getting Started with DevOps? Here are some FREE resources that might help.

Prashant Lakhera — Mon, 20 Jul 2026 00:45:49 +0000

Over the last few years, I've created several free resources to help engineers learn DevOps, improve their practical skills, and prepare for interviews. If you're starting your DevOps journey or want to strengthen your fundamentals, I hope you find these useful.
📌 Linux Performance Tuning
https://www.youtube.com/playlist?list=PLckUzKjgYDgajJVYOjNztS6Q4SOho0RKY
☁️ AWS Boto3
https://www.youtube.com/playlist?list=PLckUzKjgYDgbVbrLPpPl_MsDHQnKovLuH
💼 100 Days of DevOps Interview
https://www.youtube.com/playlist?list=PLckUzKjgYDgZ4AE6D-fKs1Z0gWCHBrk65
🐍 Python for DevOps
https://www.youtube.com/playlist?list=PLckUzKjgYDgaMCzGIvdcyOlcUTx1sBBtR
📖 Cracking the DevOps Interview (Book)
https://pratimuniyal.gumroad.com/l/cracking-the-devops-interview
🤖 I also created DevOps Open Agent, an open-source AI-powered DevOps assistant for Kubernetes, AWS, Linux, GitHub, security scanning, cloud cost analysis, and much more.
🔗 GitHub: https://lnkd.in/giri6MF5
🎥 YouTube: https://lnkd.in/g2PbG3Hc
🎉 August enrollment is now open at IdeaWeaver AI Labs!
If you're looking for a more structured, hands-on learning experience, we have three programs starting this month:
🔵 GenAI for DevOps Engineers
👉 https://ideaweaver.ai/#courses/genai-for-devops-engineers
🟢 GenAI for Beginners
👉 https://ideaweaver.ai/#courses/genai-for-beginners
🟡 NVIDIA-Certified Associate: AI Infrastructure and Operations
👉 https://ideaweaver.ai/#courses/nvidia-certification
🚀 P.S. If you're interested in learning how AI is transforming DevOps, I'm also publishing a 100 Days of GenAI for DevOps Engineers Free series in both English and Hindi, with a new lesson released every day.
🇺🇸 English: https://www.ideaweaver.ai/purchase?product_id=6786786
🇮🇳 Hindi: https://www.ideaweaver.ai/purchase?product_id=6786785
If any of these resources help you, I'd really appreciate it if you shared them with someone who's beginning their DevOps journey. Happy learning! 🙌

🚀 Day 12 of 100 Days of GenAI for DevOps: Building Docker GPT Using LLM Fine-Tuning

Prashant Lakhera — Sun, 19 Jul 2026 01:56:34 +0000

Today we're taking the next step beyond prompt engineering and RAG; we're building our own Docker GPT using LLM Fine-Tuning.

If you've ever wondered:

📌 How can I teach an LLM to understand Docker commands and troubleshooting?

📌 How do models learn domain-specific knowledge?

📌 What exactly happens during LLM fine-tuning?

📌 When should you use fine-tuning instead of RAG?

This video answers all of these questions in a practical, beginner-friendly way.

We'll cover:

✅ What LLM Fine-Tuning is

✅ Different types of fine-tuning

✅ The complete fine-tuning workflow

✅ Building a Docker-focused GPT

Whether you're a DevOps Engineer, Platform Engineer, SRE, or someone starting your GenAI journey, this video will give you a solid understanding of how custom LLMs are built.

🎥 Watch here:

🔗 English: https://youtu.be/5IoB0mfVCuA

🔗 Hindi: https://youtu.be/LSkRd_PC0YU

📚 Follow the complete 100 Days of GenAI for DevOps series:

🔗Day 12 (English): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-english/lectures/66332171

🔗Day 12(Hindi): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-hindi/lectures/66332172

If you find the series helpful, I'd really appreciate it if you could Like, Share, and Subscribe. It helps the content reach more engineers who want to learn AI for DevOps.

🎉 August enrollment is now open at IdeaWeaver AI Labs!

We have 3 programs starting this month, pick the one that fits where you are:

🔵 GenAI for DevOps Engineers

👉 https://ideaweaver.ai/#courses/genai-for-devops-engineers

🟢 GenAI for Beginners

👉 https://ideaweaver.ai/#courses/genai-for-beginners

🟡 NVIDIA-Certified Associate: AI Infrastructure and Operations

👉 https://ideaweaver.ai/#courses/nvidia-certification

🚀 What a week it's been for DevOps Open Agent!

Prashant Lakhera — Fri, 17 Jul 2026 16:03:26 +0000

Over the past week, we've been shipping new features, listening carefully to your feedback, and continuously improving the project. Here's what we've added:

✅ Security Scanning

Kubernetes cluster scanning with Trivy

Container image scanning with AI-powered remediation recommendations

✅ DevOps Open Agent v2 UI

A completely redesigned Operations Hub for a cleaner and more intuitive experience

✅ LLM-as-a-Judge

Verify AI recommendations with a second LLM, so you're not relying on a single model's opinion

✅ Linux Performance Debugging

Collect performance metrics over SSH and get AI-assisted analysis to quickly identify host bottlenecks

❤️ And the open-source community has been incredible!

In just the last two weeks on GitHub:

⭐ Stars: 1 → 35
🍴 Forks: 1 → 27
📥 178 repository clones
👀 1,500+ visitors

Seeing this kind of growth is incredibly motivating. It tells me that many DevOps, SRE, and Platform Engineering teams are looking for an open-source, self-hostable way to bring AI into their day-to-day workflows.

We're still at the beginning of this journey, and there are many more enterprise-ready features coming over the next few weeks. Stay tuned! 🚀

If you'd like to try it out, I'd love to hear your feedback. And if you find the project useful, please consider giving it a ⭐ on GitHub, it really helps the project reach more people.

🔗 GitHub: https://github.com/ideaweaver-ai/devops-open-agent

🎥 YouTube: https://www.youtube.com/watch?v=9HjWLyoHdng

✅ Day 10: 100 Days of GenAI for DevOps: MCP for DevOps Engineers: GitHub & AWS MCP Servers✅

Prashant Lakhera — Thu, 16 Jul 2026 23:54:16 +0000

Today we're diving into one of the most important technologies in the AI ecosystem: Model Context Protocol (MCP).

If you've been wondering how AI models can securely interact with external systems like GitHub, AWS, databases, Kubernetes, or other DevOps tools, MCP is the answer.

In today's video, we cover:

✅ What is MCP and why it matters

✅ How MCP works behind the scenes

✅ Setting up and using the GitHub MCP Server

✅ Connecting to the AWS MCP Server

✅ Real-world DevOps use cases for MCP

I'm also happy to share that DevOps Open Agent supports MCP. 🎉

🔗 https://github.com/ideaweaver-ai/devops-open-agent

🎥 https://youtu.be/9HjWLyoHdng

With MCP support, DevOps Open Agent can connect to external MCP servers and use them as tools, making it easier to investigate infrastructure, interact with GitHub repositories, access cloud resources, and integrate with the growing MCP ecosystem.

As more platforms adopt MCP, this becomes a powerful way to extend AI agents without building custom integrations for every service.

🔗 Day 10 (English): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-english/lectures/66317915

🔗 Day 10(Hindi): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-hindi/lectures/66317916

🚀 DevOps Open Agent v2 is Here!

Prashant Lakhera — Thu, 16 Jul 2026 00:33:47 +0000

I’m excited to introduce DevOps Open Agent v2! 🎉

Over the last few weeks, we've been working hard to make it one of the most comprehensive open-source AI platforms for DevOps engineers. Here's what's new:

🤖 New AI Agents

🔍 Performance Debugging
Debug Linux host performance over passwordless SSH (CPU, memory, disk, and network) with AI-powered root cause analysis.

🛡️ Security Scanning
Scan container images and Kubernetes clusters using Trivy, with AI-prioritized remediation recommendations.

💰 Cloud Cost Detector
Identify unused and underutilized AWS resources to optimize cloud spending.

📝 GitHub PR Reviewer
AI-powered DevOps code reviews for GitHub Pull Requests.

🔌 New Integrations

💬 Slack
Deliver AI recommendations directly to your Slack channels.

👥 Microsoft Teams
Receive the same intelligent alerts in your Teams workflows.

🚨 PagerDuty
Automatically trigger on-call incidents from AI findings with configurable cooldowns.

🔗 MCP (Model Context Protocol)
Connect to external tool servers with Ask MCP and investigation enrichment.

🧠 Qdrant (RAG)
Vector-based investigation memory for richer, context-aware AI analysis.

🧠 New LLM Providers

✨ Google Gemini
Now supported alongside OpenAI, Anthropic, OpenRouter, and Ollama.

⚡ Platform Improvements

⏰ Proactive Schedules
Run recurring Kubernetes investigations with automated AI diagnosis.

✅ LLM-as-a-Judge
Cross-check Kubernetes debugging recommendations using a second LLM.

📚 Investigations History
Browse and revisit previous investigations across all agents.

🔔 Per-Agent Notification Control
Choose exactly which of the 6 AI agents send notifications to each integration.

☁️ AWS Lambda & S3 Support
Expanded AWS coverage beyond EC2 and VPC.

🎨 Redesigned Operations Hub UI
A cleaner, unified experience across agents and integrations.

🌍 Everything is open source, self-hostable, and vendor-neutral.

Run it locally with Docker Compose and connect it to your preferred LLM provider.

⭐ GitHub: https://github.com/ideaweaver-ai/devops-open-agent

I'd love to hear your feedback, feature requests, or contributions. If you find the project useful, please consider giving it a ⭐ on GitHub, it really helps the project grow!

🎥 YouTube link: https://youtu.be/9HjWLyoHdng

🎉 August enrollment is now open at IdeaWeaver AI Labs!

We have 3 programs starting this month, pick the one that fits where you are:

🔵 GenAI for DevOps Engineers

👉 https://ideaweaver.ai/#courses/genai-for-devops-engineers

🟢 GenAI for Beginners

👉 https://ideaweaver.ai/#courses/genai-for-beginners

🟡 NVIDIA-Certified Associate: AI Infrastructure and Operations

👉 https://ideaweaver.ai/#courses/nvidia-certification

🤔 Can You Really Trust a Single LLM in Production? Just Added LLM-as-a-Judge to DevOps Open Agent 🛠️

Prashant Lakhera — Wed, 15 Jul 2026 16:07:43 +0000

🚀 One of my favorite parts of building in the open is the feedback I receive from the community.
A few days ago, someone asked me a simple but important question: If DevOps Open Agent relies on one LLM for its recommendations, how do we know those recommendations are actually correct? Shouldn't another model verify them?"
That was a great point.
Today I've added LLM-as-a-Judge to DevOps Open Agent.
💡 Instead of blindly trusting the first AI response, you can now have a second, independent LLM review the investigation.
Here's what happens:
🔍 The primary LLM analyzes the Kubernetes issue and generates:
• Root cause
• Recommended fix
• Suggested kubectl commands
• Confidence score
🧑‍⚖️ A second LLM then receives both:
• The complete investigation evidence
• The first model's diagnosis
It independently checks:
✅ Is the diagnosis supported by the evidence?
✅ Are the recommended commands safe?
✅ Was any important evidence overlooked?
✅ Is the suggested solution complete and actionable?
Finally, it returns a verdict:
🟢 Agree
🟡 Partially Agree
🔴 Disagree
One thing I want to emphasize: the judge is advisory, not authoritative.
The goal isn't to replace engineers with another AI. The goal is to give engineers another perspective before making changes to production systems.
Even better, you're free to mix providers.
➡️ Diagnose with OpenAI and validate with Anthropic.
➡️ Diagnose with Anthropic and validate with Gemini.
➡️ Use whichever combination makes the most sense for your environment.
Everything can be configured directly from the UI, or you can define default providers using environment variables.
I genuinely believe that if we're going to use AI in production operations, trust has to be built, not assumed. Features like this are a small step toward making AI-assisted DevOps more reliable.
If you're interested in AI for DevOps, Kubernetes, and SRE, I'd love your feedback.
⭐ If you find the project useful, please consider giving it a star on GitHub.
GitHub: https://lnkd.in/giri6MF5

🎉 August enrollment is now open at IdeaWeaver AI Labs!
We have 3 programs starting this month, pick the one that fits where you are:
🔵 GenAI for DevOps Engineers
👉 https://lnkd.in/g8EUki-E
🟢 GenAI for Beginners
👉 https://lnkd.in/gV3byVfh
🟡 NVIDIA-Certified Associate: AI Infrastructure and Operations
👉 https://lnkd.in/g-ByqQF4
Multiple batch timings available: weekday and weekend, morning and evening.
🇮🇳 India folks: UPI/Paytm payment option available on the enrollment page.

🚀 Day 9 of 100 Days of GenAI for DevOps: DevOps Interview Buddy using n8n, Ollama, and Gemma

Prashant Lakhera — Wed, 15 Jul 2026 03:36:13 +0000

Preparing for a DevOps interview isn't just about memorizing questions—it's about understanding concepts and improving through feedback.

In today's project, we build a DevOps Interview Buddy that runs completely locally using n8n, Ollama, and Google Gemma.

With this AI-powered assistant, you can:

✅ Practice role-based DevOps interviews
✅ Get real-time feedback on your answers
✅ Cover topics like Linux, Docker, Kubernetes, CI/CD, Cloud, Networking, and more
✅ Keep everything private by running the LLM locally with Ollama

This project demonstrates how AI agents can solve real-world problems beyond simple chatbots.

🎥 Day 9 (English): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-english/lectures/66300994

🎥 Day 9(Hindi): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-hindi/lectures/66300993

🎉 August enrollment is now open at IdeaWeaver AI Labs!

We have 3 programs starting this month, pick the one that fits where you are:

🔵 GenAI for DevOps Engineers

👉 https://ideaweaver.ai/#courses/genai-for-devops-engineers

🟢 GenAI for Beginners

👉 https://ideaweaver.ai/#courses/genai-for-beginners

🟡 NVIDIA-Certified Associate: AI Infrastructure and Operations

👉 https://ideaweaver.ai/#courses/nvidia-certification

Multiple batch timings available: weekday and weekend, morning and evening.

🔥 DevOps Open Agent now supports Linux Performance Debugging🔥

Prashant Lakhera — Tue, 14 Jul 2026 17:35:49 +0000

I'm more than happy to announce a feature I've been working on for quite some time.
DevOps Open Agent now supports Performance Debugging.
When production is slow, the hard part isn't usually is something wrong?, it's which process, on which host, and why.

With the new Performance Debugging agent, you can:
✅ Enter a hostname (or user@host)
✅ Upload a list of hosts from a file
✅ Click Start Debugging
✅ Collect Linux CPU, memory, disk, and network signals over SSH
✅ Get shared AI analysis with clear findings and recommended next steps

It's designed for how DevOps and SRE teams actually work, open source, self-hostable, and focused on actionable insight (not auto-remediation).
NOTE: You'll need passwordless SSH set up from the machine running DevOps Open Agent before you start.
If you've been waiting for host-level performance debugging alongside Kubernetes, AWS, cloud cost, and PR review, this one's for you.

👉 Repo: https://github.com/ideaweaver-ai/devops-open-agent
If you find the project useful, don't forget to ⭐ star the repo, it really helps us keep building in the open.

🚀 Enrollments for our August batch are now open! Check out the course details and register here:
https://ideaweaver.ai/#courses

🎉 August enrollment is now open - 3 programs to choose from!

Prashant Lakhera — Tue, 14 Jul 2026 05:38:50 +0000

We are now enrolling for the month of August across three programs:
🟢 GenAI for Beginners
New to Generative AI? This is your starting point. Build a solid foundation in LLMs, prompt engineering, LangChain, Hugging Face, and Ollama, then progress to building real AI agents.
🔵 GenAI for DevOps Engineers
For DevOps, Platform, and SRE engineers ready to integrate AI into their daily workflows. Covers AWS Bedrock, Kubernetes AI ops, DevOps agents, RAG, MCP, and fine-tuning, all hands-on via Google Colab.
🟡 NVIDIA-Certified Associate: AI Infrastructure and Operations (NCA-AIIO)
A 2-day intensive prep course for the NVIDIA certification. Covers GPU architecture, AI infrastructure design, cluster operations, GPU monitoring, and cloud vs on-prem trade-offs.
All programs are live, instructor-led, and cohort-based. Multiple batch timings available: weekday and weekend, morning and evening to fit your schedule.
👉 Full details and enrollment: https://ideaweaver.ai/#courses
🇮🇳 India folks, you can enroll via Paytm/UPI (₹4,500) by emailing your transaction ID to help@ideaweaver.ai.

🚀 Day 8 of 100 Days of GenAI for DevOps: Introduction to AI Agents

Prashant Lakhera — Mon, 13 Jul 2026 15:59:05 +0000

If you've been hearing terms like Agentic AI, Function Calling, Tool Calling, MCP, CrewAI, LangChain, LangGraph, or ReAct and wondering how they all fit together, this video is for you.

In this session, we cover:

✅ What an AI Agent actually is
✅ How AI Agents are different from a traditional LLM
✅ Why function calling changed everything
✅ How LLMs decide which tool to use
✅ The ReAct framework (Reason + Act)
✅ Different types of AI agents and agent architectures
✅ Popular frameworks like CrewAI, LangChain, LangGraph, and more

This video is designed for DevOps, SRE, Platform, and Cloud Engineers who want to understand AI Agents from the ground up before building real-world agentic applications.

Whether you're just getting started or planning to build your own AI-powered automation, this lesson will give you the foundation you need.

📺 Watch the video here:

Day 8 (English): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-english/lectures/66279569

Day 8 (Hindi): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-hindi/lectures/66279588

🎓 Want Live Instructor-Led Training?

https://ideaweaver.ai/#courses

➡️ If you're looking to try an AI agent built specifically for DevOps, check out DevOps Open Agent: https://github.com/ideaweaver-ai/devops-open-agent

➡️ IdeaWeaver Small Language Builder: https://github.com/ideaweaver-ai/ideaweaver-slm-builder

If you're following the 100 Days of GenAI for DevOps series, don't forget to like, share, and follow along. We still have many exciting projects and hands-on labs coming up!

🚀 Welcome to Day 7 of 100 Days of GenAI for DevOps!

Prashant Lakhera — Sun, 12 Jul 2026 05:49:00 +0000

Over the past few days, we've been learning the fundamentals of Generative AI and building practical projects that solve real DevOps problems.
Today, we're taking another step forward with a project that almost every DevOps engineer, SRE, and Platform Engineer can relate to.
🔍 Apache Log Analyzer using LangChain and OpenAI
Server logs contain a large amount of information, but manually reviewing thousands of log entries to identify issues is time-consuming and often frustrating.
In this project, we'll build an AI-powered log analyzer that can understand Apache access logs and explain what's happening in plain English.
In this session, you'll learn how to:
✅ Read and process Apache log files using Python
✅ Use LangChain to build an AI-powered analysis pipeline
✅ Send log data to OpenAI for intelligent reasoning
✅ Identify common issues such as 404 errors, failed requests, suspicious URLs, and abnormal traffic patterns
✅ Receive human-readable explanations instead of manually interpreting raw log entries
✅ Extend the same approach to analyze other infrastructure and application logs
This is a practical example of how Generative AI can help DevOps engineers spend less time reading logs and more time solving problems.
If you're a DevOps Engineer, SRE, Platform Engineer, or anyone interested in AI for Infrastructure, this project is a great introduction to building intelligent troubleshooting tools.
🎓 Want Live Instructor-Led Training?
If you'd like hands-on, instructor-led sessions where you can ask questions and get your doubts resolved live, check out our courses:
👉 https://ideaweaver.ai/#courses
📚 Follow Along Daily (Registration Required)
Day 7 (English): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-english/lectures/66276273
Day 7 (Hindi): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-hindi/lectures/66276274
➡️ If you're looking to try an AI agent built specifically for DevOps, check out DevOps Open Agent:
https://github.com/ideaweaver-ai/devops-open-agent
➡️ IdeaWeaver Small Language Builder: https://github.com/ideaweaver-ai/ideaweaver-slm-builder
➡️ GitHub: https://github.com/ideaweaver-ai/From-Software-DevOps-Engineer-Generative-AI-Engineer/blob/main/llm_log_analysis/apache_log_analyzer_langchain.py