DEV Community: Prashant Lakhera

📌 Calling an LLM API Doesn’t Mean You Understand How an LLM Works

Prashant Lakhera — Sat, 29 Aug 2026 00:12:55 +0000

One common mistake I see among people entering the AI field:

They make an API call to an LLM, get a response, build a small application, and think they understand how LLMs work.

But that’s only the surface.

The real learning starts when you begin asking:

→ What is actually happening inside a neural network?

→ Why are GPUs so important for AI?

→ How does a Transformer work?

→ What are attention and tokenization?

→ How does training actually happen?

→ How do you build and train a language model yourself?

These are also the kinds of fundamentals that can separate you in a serious AI/GenAI interview. That’s exactly why I wrote:

📕 Building A Small Language Model from Scratch: A Practical Guide

It’s an 854-page practical journey that starts with the fundamentals: Neural Networks, GPUs, Transformers, tokenization, and attention, and gradually takes you toward building your own language model from scratch.

If you want to go deeper than simply calling an API, the book is still available:

📘 Gumroad: https://plakhera.gumroad.com/l/BuildingASmallLanguageModelfromScratch

📘 Leanpub: https://leanpub.com/buildingasmalllanguagemodelfromscratch

📘 Amazon: https://www.amazon.com/Building-Small-Language-Model-Scratch-ebook/dp/B0G64SQ4F8/

🚀 Looking for a structured interview-preparation path?

I’m also running a 3-month program:

Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers

The program focuses on:

→ Generative AI & LLM Engineering

→ Python & DSA

→ System Design

→ DevOps/SRE Automation with AI

→ Hands-on AI projects

→ Interview-focused preparation

📌 Morning Batch: https://www.ideaweaver.ai/purchase?product_id=6827463

📌 Evening Batch: https://www.ideaweaver.ai/purchase?product_id=6827464

📌 Self-Paced Batch: https://www.ideaweaver.ai/purchase?product_id=6827466

📌 Complete Program Details: https://www.ideaweaver.ai/courses/cracking-the-genai-interview-for-devops-sre-platform-forward-deployed-engineers-morning1/lectures/66591133

Don’t just learn how to call an LLM. Learn what is happening underneath it.

🎥 Cracking the AWS Cloud Support Engineer Interview: Recording Now Available🎥

Prashant Lakhera — Thu, 27 Aug 2026 16:03:14 +0000

Thank you, everyone, for joining yesterday’s session on Cracking the AWS Cloud Support Engineer Interview.

We discussed how to approach real-world AWS troubleshooting scenarios, structure your answers during an interview, and think like a Cloud Support Engineer rather than simply memorizing AWS services.

🎥 The recording of yesterday’s session is now available.

https://youtu.be/g8Cgg09PAEs

If you’re looking for more structured interview preparation, our 3-month program — Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers starts September 14.

The program focuses on:

✔️ GenAI for DevOps & Infrastructure Engineers
✔️ DevOps/SRE scenario-based interview questions
✔️ Python, DSA & LeetCode
✔️ System Design
✔️ AI/LLM infrastructure
✔️ Interview preparation for companies such as OpenAI, Anthropic, NVIDIA, AWS, Google, Meta, Microsoft, Apple, Tesla & SpaceX

Batches (PST):

Morning: Mon–Fri, 7–8 AM (Thursday off)
Evening: Mon–Fri, 7–8 PM (Thursday off)
Weekend: Sat–Sun, 7–9 AM

Pricing:

Live Cohort — $299 / ₹24,999
Self-Paced — $149 / ₹12,999

📌 Morning Batch: https://www.ideaweaver.ai/purchase?product_id=6827463

📌 Evening Batch: https://www.ideaweaver.ai/purchase?product_id=6827464

📌 Self-Paced Batch: https://www.ideaweaver.ai/purchase?product_id=6827466

📌 Complete Program Details: https://www.ideaweaver.ai/courses/cracking-the-genai-interview-for-devops-sre-platform-forward-deployed-engineers-morning1/lectures/66591133

📌 30 Most-Asked AWS Cloud Support Engineer Interview Questions: Free WEBINAR TONIGHT • 7 PM PST📌

Prashant Lakhera — Wed, 26 Aug 2026 17:54:06 +0000

30 Most-Asked Questions for the AWS Cloud Support Engineer Interview

If you’re preparing for an AWS Cloud Support Engineer role, I’ve put together 30 of the most commonly asked interview questions covering AWS, Linux, networking, troubleshooting, security, monitoring, and real-world production scenarios.

I’m also conducting a FREE live session to help you prepare:

✔️ Date: Aug 26

✔️ Time: 7–8 PM PST
🔗 Joining Link: https://www.linkedin.com/events/7497868527895379968?viewAsMember=true

If you’re looking for a more structured interview-preparation program, our 3-month Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers program starts Sep 14.

🔗 Morning Batch:
https://www.ideaweaver.ai/purchase?product_id=6827463

🔗 Evening Batch:
https://www.ideaweaver.ai/purchase?product_id=6827464

🔗 Self-Paced Batch:
https://www.ideaweaver.ai/purchase?product_id=6827466

🔗 Complete Program Details:
https://www.ideaweaver.ai/courses/cracking-the-genai-interview-for-devops-sre-platform-forward-deployed-engineers-morning1/lectures/66591133

📌 MCP for DevOps: GitHub and AWS MCP Server 📌

Prashant Lakhera — Wed, 26 Aug 2026 15:30:13 +0000

There are plenty of videos and tutorials explaining MCP (Model Context Protocol).

But not many explain MCP from a DevOps engineer’s point of view.

So here’s my attempt to make MCP practical for DevOps using two real-world MCP servers:

1️⃣ GitHub MCP Server: connecting AI with repositories, issues, pull requests, and developer workflows

2️⃣ AWS MCP Server: connecting AI with AWS services and cloud operations

The goal is simple: understand MCP by using it for real DevOps workflows, not just theory.

🎥 Watch here:

English: https://youtu.be/pRs8orB1Ee8?si=NC3D18kVZ-m7v4Xm

Hindi: https://youtu.be/ycadAzGFy1Y

And if you want to go deeper into GenAI, LLMs, AI Agents, MCP, RAG, Python, System Design, and AI-powered DevOps, our 3-month program starts September 14.

📌 Morning Batch: https://www.ideaweaver.ai/purchase?product_id=6827463
📌 Evening Batch: https://www.ideaweaver.ai/purchase?product_id=6827464
📌 Self-Paced Batch: https://www.ideaweaver.ai/purchase?product_id=6827466
📌 Complete Program Details: https://www.ideaweaver.ai/courses/cracking-the-genai-interview-for-devops-sre-platform-forward-deployed-engineers-morning1/lectures/66591133

📌 Cracking the DevOps Interview: Start Your Preparation for Free📌

Prashant Lakhera — Sun, 23 Aug 2026 02:17:14 +0000

If you are preparing for a DevOps interview and don't know where to start, here is something that might help.
📌 I have made my book, Cracking the DevOps Interview, completely FREE.
You can use it to practice interview questions, refresh important DevOps concepts, and identify areas where you need more preparation.
🔗 Free book:
https://pratimuniyal.gumroad.com/l/cracking-the-devops-interview
But there is something important I have realized while preparing engineers for interviews:
The interview landscape is changing.
Today, companies are not only looking for engineers who understand Linux, AWS, Kubernetes, Terraform, CI/CD, and troubleshooting.
For many modern infrastructure roles, engineers are increasingly expected to understand:
✔️ Generative AI & LLM fundamentals
✔️ Python & DSA problem solving
✔️ System Design
✔️ AI Agents & automation
✔️ GPU and AI infrastructure
✔️ DevOps/SRE troubleshooting
How AI can be applied to real infrastructure problems
That is exactly why I created the 90-Day Intensive Program:
Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers
The goal is simple:
Spend 90 days preparing consistently, building the right skills, solving interview-style problems, and becoming better prepared for the next generation of infrastructure engineering roles.
📌 Live Batch Schedule - PST
Morning: Mon–Fri, 7–8 AM (Thursday off)
Evening: Mon–Fri, 7–8 PM (Thursday off)
Weekend: Sat–Sun, 7–9 AM
📌 Pricing
Live Cohort: $299 / ₹24,999
Self-Paced: $149 / ₹12,999
📌 Morning Batch:
https://www.ideaweaver.ai/purchase?product_id=6827463
📌 Evening Batch:
https://www.ideaweaver.ai/purchase?product_id=6827464
📌 Self-Paced:
https://www.ideaweaver.ai/purchase?product_id=6827466
📌 Complete Program Details:
https://www.ideaweaver.ai
If you are not ready for the program yet, start with the free book.
Start small. Learn consistently. Solve one problem at a time.
You don't need to know everything before you start. You just need to start.

📌 Understanding LLM Inference: Prefill, Decode, and KV Cache📌

Prashant Lakhera — Sat, 22 Aug 2026 17:04:11 +0000

Most people use ChatGPT every day. But have you ever wondered what actually happens after you press Enter? The answer does not magically appear all at once.

Behind the scenes, an LLM undergoes a complex inference process.

And if you are learning LLM Engineering, AI Infrastructure, GPU Infrastructure, or GenAI for DevOps, there are three concepts you should understand:

📌 Prefill — The model processes your complete input prompt.

📌 Decode — The model starts generating the response, one token at a time.

📌 KV Cache — Instead of recalculating information from all previous tokens again and again, the model remembers useful computations and reuses them.

That last part is especially important. KV Cache can make inference much faster.

But there is a trade-off: KV Cache consumes GPU memory.

And once you start serving hundreds or thousands of users, managing that GPU memory efficiently becomes a serious infrastructure problem.

This is also why inference frameworks such as vLLM and SGLang become important.

If you are coming from a DevOps, SRE, Platform Engineering, Cloud, or Forward-Deployed Engineering background and trying to move deeper into GenAI infrastructure, this is one of those foundational concepts worth understanding.

🎥 Video:
https://www.youtube.com/watch?v=05u-bgl3ETg

I’m also starting a 90-Day Intensive Program, Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers in September.

We’ll go much deeper into:

📌 Generative AI & LLM Engineering

📌 Python & DSA

📌 System Design

📌 DevOps & SRE Automation

📌 Hands-on AI Projects

📌Morning Batch: https://www.ideaweaver.ai/purchase?product_id=6827463

📌Evening Batch: https://www.ideaweaver.ai/purchase?product_id=6827464

📌Self-paced Batch: https://www.ideaweaver.ai/purchase?product_id=6827466

Image Ref: https://developer.nvidia.com/blog/mastering-llm-techniques-inference-optimization/

🚀 Multi-GPU Inference: explained simply 🚀

Prashant Lakhera — Fri, 21 Aug 2026 15:20:12 +0000

When people first hear multiple GPUs, it’s easy to think:

More GPUs = faster LLM.

But that’s not always the case.

The real question is:

Why do we need multiple GPUs in the first place?

There are mainly two problems:

📌 The model fits on one GPU, but we need to serve more users.

📌 The model is too large to fit on a single GPU.

And that leads us to three important techniques:

1️⃣ Data Parallelism → Copy the model

Run a complete copy of the same model on multiple GPUs and distribute incoming requests among them.

Goal: Higher throughput

2️⃣ Tensor Parallelism → Split the layer

Split computations inside a model layer across multiple GPUs.

Because the GPUs need to communicate frequently, high-speed GPU interconnects such as NVLink and NVSwitch become very important.

Goal: Run larger models and parallelize computation

3️⃣ Pipeline Parallelism → Split the layers

Different groups of model layers run on different GPUs.

For example:

GPU 0 → Layers 1–4

GPU 1 → Layers 5–8

GPU 2 → Layers 9–12

The output of one stage is passed to the next, almost like an assembly line.

Goal: Scale very large models across GPUs and nodes

The easiest way to remember all three:

Data Parallelism → Copy the model

Tensor Parallelism → Split the layer

Pipeline Parallelism → Split the layers

And for very large deployments, these techniques can also be combined using Hybrid Parallelism.

📌 Want to learn these concepts from an interview perspective?

I’m starting a 90-Day Intensive Program:

Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers

We’ll connect traditional infrastructure engineering with the skills increasingly important for modern AI infrastructure roles:

✔️ Generative AI & LLM Engineering

✔️ GPU & AI Infrastructure

✔️ Kubernetes for AI workloads

✔️ Python & DSA

✔️ System Design

✔️ DevOps & SRE

✔️ Interview-style troubleshooting scenarios

The goal is not just to learn AI terminology, but to understand how these systems actually work and how to explain and troubleshoot them during interviews.

📌 Morning Batch: https://lnkd.in/gEGWguSE

📌 Evening Batch: https://lnkd.in/gdqXHS_g

📌 Self-paced Batch: https://lnkd.in/gVTSSZZH

🎥 YouTube video

English: https://lnkd.in/gmUnipkh

Hindi: https://lnkd.in/giCH6BNP

📌 The TAO Loop: How AI Agents Actually Work📌

Prashant Lakhera — Thu, 20 Aug 2026 15:56:09 +0000

Many people jump directly into building AI agents using frameworks like LangGraph, CrewAI, or other agent frameworks.

But before building an agent, it’s important to understand how an agent actually works under the hood.

One of the fundamental concepts to understand is the TAO Loop:

Thought → Action → Observation → Repeat

In this blog, I explain the TAO Loop in beginner-friendly language, with diagrams and a practical Kubernetes troubleshooting example showing how an agent thinks, uses tools, observes the results, and decides what to do next.

📌 Prefer video? I’ve also explained the concept here:

https://youtu.be/Ng1z3TTAghY?si=9fSSZoah6VGCoqo3

I’ll be covering concepts like AI Agents, Tool Calling, ReAct, MCP, GenAI, GPU Infrastructure, Python, System Design, and DevOps/SRE automation in my upcoming 90-Day Intensive Program:

Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers

📌 Morning Batch:
https://www.ideaweaver.ai/purchase?product_id=6827463

📌 Evening Batch:
https://www.ideaweaver.ai/purchase?product_id=6827464

📌 Self-Paced:
https://www.ideaweaver.ai/purchase?product_id=6827466

🚀 The DevOps, SRE, Platform Engineering, and Forward Deploy Engineer interview landscape is changing

Prashant Lakhera — Thu, 20 Aug 2026 01:28:17 +0000

For years, if you were preparing for a DevOps, SRE, or Platform Engineering or Forward Deploy Engineer interview, the roadmap was pretty clear:

✅ Linux

✅ Networking

✅ Cloud

✅ Docker

✅ Kubernetes

✅ Terraform

✅ CI/CD

✅ On-call & troubleshooting

And these skills are still extremely important.

But there’s a new layer being added to the infrastructure world:

🤖 Generative AI

Today:

→ LLMs are running on Kubernetes

→ GPUs are becoming infrastructure

→ AI inference is becoming a production workload

→ AI agents are helping with incident response and automation

→ DevOps and SRE teams are starting to build and operate AI-powered systems

This means companies increasingly need engineers who understand both Infrastructure + Generative AI.

And that creates a new challenge:

How do you prepare for interviews when you need to learn AI, coding, system design, and infrastructure together?

That’s exactly why we built:

🎯 Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers

A 90-day intensive interview preparation program from IdeaWeaver AI Labs.

This is not just another GenAI course.

And it’s definitely not just about prompt engineering.

The program focuses on 5 interview tracks:

✅ Generative AI & LLM Engineering

✅ Python & DSA

✅ System Design

✅ DevOps & SRE Automation

✅ Hands-on AI Projects

And you’ll learn from three instructors with different areas of expertise:

👨‍💻 Prashant Lakhera — Ex-Red Hat & Salesforce

GenAI, LLMs & AI Infrastructure

👨‍💻 Rahul — Ex-Uber

Python, DSA & Coding Interviews

👨‍💻 Pramod — Ex-Yahoo & Apple

System Design

One structured roadmap.

One goal:

👉 Help you prepare for the next generation of infrastructure engineering interviews.

And if you’re thinking:

"I’m a DevOps/SRE engineer. I don’t have GenAI experience. Am I already too late?"

Absolutely not.

We start with the fundamentals.

You don’t need to become an AI researcher.

Your existing knowledge of Linux, Cloud, Kubernetes, networking, troubleshooting, and distributed systems is already a huge advantage.

You’re not starting from zero.

You’re simply adding the GenAI layer on top of your infrastructure experience. 🚀

📅 September 14 – December 12

🔗 Morning batch: https://www.ideaweaver.ai/purchase?product_id=6827463

🔗 Evening batch: https://www.ideaweaver.ai/purchase?product_id=6827464

🔗 Self-paced batch: https://www.ideaweaver.ai/purchase?product_id=6827466

If you’re a:

✅ DevOps Engineer

✅ Site Reliability Engineer (SRE)

✅ Platform Engineer

✅ Cloud Engineer

✅ Systems Engineer

✅ Forward-Deployed Engineer

…and you want to prepare for interviews where Infrastructure meets AI, this program is designed for you.

🎥 Video link: https://youtu.be/w8MEvjgBwfM

🔥 Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers🔥

Prashant Lakhera — Wed, 19 Aug 2026 14:51:51 +0000

Most interview prep still treats DevOps, SRE, and Platform engineers like they only need Kubernetes questions. That’s no longer the job.

Hiring teams now expect traditional infrastructure knowledge plus Generative AI, LLMs, Python, system design, automation, and AI agents.

That’s why we’re launching something we’ve never offered before:

Cracking the GenAI Interview for DevOps, SRE, Platform & Forward-Deployed Engineers

A 90-day intensive (September 14 – December 12). Live, instructor-led. 6 days a week(Thursday is the break).

This is not a theory dump. It combines five interview tracks:

✅ Generative AI & LLM Engineering

✅ Python & DSA

✅ System Design

✅ DevOps/SRE Automation

✅ Hands-on AI Projects

You’ll learn from 3 engineers:

📌 Prashant Lakhera — Ex-Red Hat & Salesforce · GenAI & LLMs

📌 Rahul — Ex-Uber · Python & DSA

📌 Pramod — Ex-Yahoo & Apple · System Design

Weekly cadence:

🗓️ Monday — GenAI for DevOps

🗓️ Tuesday — Python for DevOps / Interview Prep

🗓️ Wednesday — GenAI for DevOps Project

🗓️ Thursday — Break

🗓️ Friday — System Design

🗓️ Saturday & Sunday — Python for DSA

We start from the basics. Basic DevOps and Python knowledge is good to have, not a hard blocker. If students ask, we can adjust the content, but we will complete every concept.

Batches (all times PST):

✅ Morning: Mon–Fri, 7–8 am PST (Thu off)

✅ Evening: Mon–Fri, 7–8 pm PST (Thu off)

✅ Weekend (both batches): Sat–Sun, 7–9 am PST

Price:

💰Live cohort — $299 / ₹24,999

💰Self-paced — $149 / ₹12,999

Enroll: https://www.ideaweaver.ai

🔗 Morning: https://www.ideaweaver.ai/purchase?product_id=6827463

🔗 Evening: https://www.ideaweaver.ai/purchase?product_id=6827464

🔗 Self-paced: https://www.ideaweaver.ai/purchase?product_id=6827466

If you want to interview for roles where infrastructure and GenAI now sit in the same room , this is the program.

GenAI #DevOps #SRE #PlatformEngineering #SystemDesign #Python #DSA #InterviewPrep #IdeaWeaverAILabs

🚀 Day 18: 100 Days of GenAI for DevOps: Kubernetes & GPUs

Prashant Lakhera — Thu, 13 Aug 2026 03:02:23 +0000

How does Kubernetes actually run an LLM on a GPU?

That was the focus of Day 18.

Kubernetes understands CPU and memory by default, but GPUs require additional components. We walked through the complete journey:

🔹 NFD (Node Feature Discovery) — Discovers GPU hardware on Kubernetes nodes.

🔹 GFD (GPU Feature Discovery) — Provides detailed NVIDIA GPU information such as model, memory, architecture, and MIG capability.

🔹 NVIDIA Device Plugin — Makes GPUs available as resources that Kubernetes workloads can request.

🔹 GPU Scheduling — We explored how nodeSelector, Node Affinity, and Taints/Tolerations help us choose the right GPU and protect expensive GPU nodes.

🔹 Dynamic Resource Allocation (DRA) — Instead of simply saying “Give me one GPU,” DRA allows workloads to describe what kind of device they need. We also looked at how DRA has evolved in Kubernetes 1.36.

🔹 NVIDIA GPU Operator — Automates much of the NVIDIA GPU software stack, including drivers, GPU discovery, the device plugin, container support, MIG management, and monitoring.

🔹 GPU Sharing — We compared Time Slicing and MIG and discussed when sharing an expensive GPU makes sense.

🔹 GPU Monitoring — We looked at how DCGM Exporter + Prometheus + Grafana can provide visibility into GPU utilization, memory, temperature, power, and health.

The big picture is:

Physical GPU → Discover → Expose → Schedule → Allocate → Run LLM → Monitor

For DevOps, SRE, and Platform Engineers moving into GenAI, GPUs are becoming another critical infrastructure resource we need to understand and manage.

Day 18/100 completed! 🚀

🚀 Want to learn GenAI from a DevOps Engineer's perspective?

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

📚 Follow Along Daily (Registration Required)

🎥 Day 18 (English): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-english/lectures/66538878

🎥 Day 18(Hindi): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-hindi/lectures/66538879

🚀 Day 17 of 100 Days of GenAI for DevOps — Understanding vLLM

Prashant Lakhera — Wed, 12 Aug 2026 02:51:40 +0000

Running an LLM on your laptop is one thing.

Serving that LLM efficiently to hundreds or thousands of users is a completely different challenge.

In Day 17 of 100 Days of GenAI for DevOps, we explore vLLM, an open-source LLM serving framework designed for fast and efficient inference.

In this session, we cover:

🔹 Why we need a dedicated LLM serving framework
🔹 What happens when multiple users send requests to an LLM
🔹 How vLLM improves LLM inference and throughput
🔹 Running an LLM using vLLM
🔹 Exposing the model through an OpenAI-compatible API
🔹 Why all of this matters for DevOps and AI infrastructure engineers

The goal of this series is not just to learn GenAI concepts.

It is to understand the infrastructure behind GenAI applications: GPUs, inference, model serving, Kubernetes, observability, scalability, and eventually how all these pieces fit together in production.

🚀 Want to learn GenAI from a DevOps Engineer's perspective?
📅 New batch starting soon!

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

🎥 Day 17 (English): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-english/lectures/66527584

Day 17(Hindi): https://www.ideaweaver.ai/courses/100-days-of-genai-for-devops-hindi/lectures/66527587

If you're a DevOps, SRE, Platform, Cloud, or Infrastructure Engineer trying to understand how the GenAI infrastructure stack actually works, I hope this series helps you along the way.