Rahul Arakeri

As a Sr Principal Technologist and Bar Raiser at Amazon who's conducted hundreds of interviews, I've seen how difficult getting quality interview practice can be. Now, AI tools are changing the game. ## Amazon Interviews: Hard Yet Predictable Amazon interviews present a mind-bogglingly high bar: you must demonstrate you're better than 50% of current employees at your level. However, the structure follows a consistent pattern centered on Leadership Principles (LPs). https://lnkd.in/e7Dd8PHG Each interviewer typically explores 3 LPs in an hour, probing deeply into your experience—what you did, why you chose that approach, challenges faced, and lessons learned. For technical roles, expect additional focus on domain expertise, coding skills, and fundamentals relevant to your specialty. ## Using AI as Your Interview Coach Here's my recommended approach: 1. **Gather your materials as text files:**   - The specific job description   - Your resume   - Amazon's Leadership Principles 2. **Create a practice environment:**   Upload these text files to an AI assistant (Claude, ChatGPT, etc.) with a prompt like this (the part about "one by one" is crucial):   *"Act as an interview coach specializing in Amazon interviews. Read my resume, job description, and Amazon's Leadership Principles. Ask me behavioral questions one by one, probing into my experience and each LP. Include follow-up questions focusing on how/why I did things, challenges, and learnings. Conclude with STAR method assessment and improvement suggestions."* 3. **Practice strategically:**   - Focus on telling concise stories with meaningful metrics   - Get comfortable with the depth of follow-up questions   - Use AI feedback to refine your examples and delivery   - Utilize voice interfaces available in some AI tools to practice speaking about your experiences out loud—this builds verbal fluency crucial for the actual interview ## Why This Works What makes this approach effective is the unlimited practice and structured feedback without the cost of a coach. The AI won't get tired of asking you to elaborate or challenge your thinking—exactly what Amazon interviewers do. Also, every interaction with those tools is unique and it won't get repetitive. By simulating the intense questioning style and receiving feedback through the STAR framework (Situation, Task, Action, Results), you'll develop the muscle memory needed to navigate the real interview confidently. Remember to use these tools not just for rehearsing answers, but for genuinely reflecting on your experiences through the lens of Amazon's culture. The best candidates demonstrate authentic alignment with Leadership Principles, not memorized responses. Have you tried AI for interview prep? I'd love to hear your experiences in the comments.

350

18 Comments

$100K H-1B visa fee: A challenge or an opportunity for India? With the recent order imposing a $100,000 fee on H-1B visas, many predict tough times for Indian tech talent in the US. But here’s another way to see it — this could accelerate a shift we were already witnessing: global companies investing more in India. We already have a strong talent pipeline and competitive salaries. The next step is clear — we need to create the right environment and culture to innovate. That means: - Expanding beyond metros into Tier-2 cities. - Building ecosystems where engineers don’t just deliver, but invent. - Solving infrastructure and workplace challenges that hold back creativity. - Taking the best from Western corporate culture (clarity, meritocracy, ownership) and blending it with India’s sense of belonging, community, and “individual first → team first” mindset. This is not just about support work moving here. It’s about India stepping into a leadership role in global innovation. I truly believe — this century belongs to India. 💡 What do you think — is this policy a setback, or a catalyst for India to lead? 🔗 Source: https://lnkd.in/d6DFUQut #H1B #TechInIndia #Innovation #Bharat

29

8 Comments

𝐆𝐨𝐨𝐠𝐥𝐞 𝐭𝐨 𝐈𝐧𝐯𝐞𝐬𝐭 $𝟏𝟓 𝐁𝐧 𝐢𝐧 𝐀𝐈 𝐃𝐚𝐭𝐚 𝐂𝐞𝐧𝐭𝐫𝐞 𝐢𝐧 𝐕𝐢𝐳𝐚𝐠 Google Cloud CEO Thomas Kurian announced that the tech giant will invest $15 billion to build the world’s largest AI and data centre in Visakhapatnam, Andhra Pradesh. This hub will be Google’s biggest AI facility outside the US and will feature Tensor Processing Units (TPUs) for cutting-edge AI computing. “This gigawatt-scale AI hub represents a major step in supporting India’s digital infrastructure goals, and we’re grateful for the support of the Government of India and the state of Andhra Pradesh,” said Kurian. Google Cloud CEO Kurian made this announcement during Google’s event on “India’s AI-Powered Future and Advancing the Vision of Viksit Bharat 2047,” in the presence of Ashwini Vaishnaw, Minister of Electronics & IT; Nirmala Sitharaman, Finance Minister; and N. Chandrababu Naidu, Chief Minister of Andhra Pradesh. Google Cloud Thomas Kurian Ministry of Electronics and Information Technology Via: Outlook Business | Deepsekhar Choudhury (Story link in the comment)

21

1 Comment

Today’s H1B announcement could prove to be one of the costliest self-goals for the U.S. — hitting hardest in the very area it must dominate to win the tech race with China: tech talent and innovation. The reality is that U.S. tech skill gaps remain a huge and persistent challenge across key roles in AI, cybersecurity, cloud, and operations. The latest #RobertHalf2025 Tech Talent Report shows the gap is widening: 87% of tech leaders struggle to hire skilled candidates 76% report gaps within their own teams, with nearly 45% citing project delays Shortages are sharpest in AI/ML & data science (44%), IT operations (39%), cybersecurity (30%), automation (25%), software/web development (22%), and cloud (20%) 69% say the problem has worsened in the past year, making security, AI adoption, modernization, and employee upskilling urgent priorities Food for thought... https://lnkd.in/gNuVZt2n

93

14 Comments

Why modern workloads need an infrastructure with a clustered control plane. Since working at Zynga, I have been trying to build a clustered infrastructure control plane for on-prem infrastructure to simplify application development. As I dug into understanding Prism Element and Prism Central's architecture, it struck me that my holy grail, building an easy-to-use and configured clustered control plane, was a solved problem. Nutanix had what I was trying to build. Why was it my holy grail? A simple principle of software engineering is that it's easier to write code on top of robust systems than on top of fragile systems. If the system is robust, then it's forgiving of programming errors. A great example is performance; your code doesn't have to be as efficient if your hardware is fast enough. If that intuition doesn't work, when you write code, you introduce bugs. It's much easier to debug the new code you just wrote than debug the code you use that you may have no access to. If that intuition doesn't work, consider a server running many applications and then installing a new one. It's much easier to debug the availability of the latest application if the server without the application has predictable availability. Traditional applications that run in a single VM rely on the OS to be more reliable than their application. So, when their application crashes, they can debug it in isolation. One of the astonishing engineering outcomes of the last twenty years was the emergence of hypervisors that are more reliable than the OS running the application. That enabled significant consolidation, engineering efficiency, and operational efficiency. The challenge with clustered applications is that they have no analog to the hypervisor that manages their infrastructure. As a result, the application cluster control plane manages the application and the infrastructure. The recent re-infatuation with bare metal is a side effect of this problem. If the virtualization layer can't offer a clustered control plane, then the application must. And if the application can do it, why do I need the virtualization layer? Virtualization provides a container on a server. That container and the server are disposable. So why am I paying money and cost? If your hypervisor doesn't have a clustered control plane, you may be right. And since I assumed Nutanix didn't have one, I planned to build one here. Imagine my surprise when I discovered that it has been in shipping for 10 years. Part of the elegance of the Nutanix VDI solution is the Prism Element's clustered control plane. Using Nutanix infrastructure, you can rely on the underlying infrastructure control plane to be more reliable than the application control plane. More importantly, you can offload some of the complexity of the application control plane. Most importantly, you can share your infrastructure easily. Welcome to the future.

46

8 Comments

One uncomfortable thought that terrifies me lately is that in the day and age of AI-assisted coding the concept of System Design will change again. Hear me out. SysDesign used to be about patterns. When the "bad OOP" [1] era was in full bloom, a "good" "system designer" effectively was the person who can tell you where and how to use a SQL database and how to [not] use the visitor and the observer patterns. Throw in some knowledge of how to separate the read path from the write path, and how to invalidate caches, and one is a decent engineer SysDesign-wise as of ~2015. If the candidate can clearly articulate how strong consistency is different from eventual consistency, and which cloud primitives are best at replacing which self-hosted solutions, they truly are on top of their game. As the "bad OOP" [1] era started its decline, by ~2020, we had a period of renaissance for System Design. People started talking about causality, vector clocks, logical sharding, read-your-own-writes, consistency, message queues vs. message buses, idempotency and fencing tokens. I remember in ~2022 you could base an interview on offering the candidate to explain how they understand the CAP theorem after they brought it up in the context of consistency models. For a well-articulating candidate, asking how the Bitcoin protocol maps to these "C, "A", and "P" letters was almost guaranteed to result in an insightful conversation, with tons of signal to extract from follow-up questions. Now we are entering the post-OOP era, which is great. Rust, for example, specifically makes the claim that we do not need inheritance and what not for the language to be powerful and easy to use. Fantastic. Unfortunately, humankind entering the post-OOP era coincided with, well, Cursor. So we are at this weird spot where on the one hand it still is seemingly important to understand SysDesign fundamentals. On the other hand, an increasing share of code is being written with or by the AI assistant. All in all, using tools to boost productivity is great. On the flip side, AI-assisted coding makes it far too easy for an inexperienced developer to make expensive mistakes reliability- and consistency-wise when it comes to making changes to a complex system with many moving parts. Naturally, on this journey towards conquering complexity, we're back to square one: SysDesign is about making sure the candidate is _less likely_ to use the AI in a way that introduces an expensive-to-debug error. Which, in my book, is even worse than when a SysDesign interview is about "tell me about the pros and cons of the visitor pattern". Because instead of designing systems thoughtfully we're inevitably teaching less senior engineers how to speed-run this game of "shipping something". The end result, predictably, will be far more legacy code that no one dares to touch. Only this time the people who wrote the code did not just leave. They never existed in the first place. It was LLMs all the way down.

27

4 Comments

Good interview with Jensen Huang and I could not agree more: "You’re not going to lose your job to an AI, but you’re going to lose your job to someone who uses AI." "In today's landscape, understanding how to use Al is no longer a luxury; it's a critical survival skill." "Mark Cuban's investment in free Al bootcamps for underserved high school students exemplifies this philosophy. Cuban urges learners to embrace Al early, absorbing its principles and applications to remain competitive in the evolving job market."

45

5 Comments

🚀 𝗚𝗼𝗼𝗴𝗹𝗲 𝗷𝘂𝘀𝘁 𝗿𝗲𝗹𝗲𝗮𝘀𝗲𝗱 𝗔𝗗𝗞: 𝗧𝗵𝗲 𝗔𝗴𝗲𝗻𝘁 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 𝗞𝗶𝘁 – 𝗮 𝗴𝗮𝗺𝗲-𝗰𝗵𝗮𝗻𝗴𝗲𝗿 𝗳𝗼𝗿 𝗺𝘂𝗹𝘁𝗶-𝗮𝗴𝗲𝗻𝘁 𝗔𝗜 𝘀𝘆𝘀𝘁𝗲𝗺𝘀! At Google Cloud Next 2025, Google unveiled ADK (Agent Development Kit)—an open-source framework purpose-built for developers creating production-grade multi-agent applications. This is not just another SDK—this is the framework behind Google’s own Agentspace and Customer Engagement Suite (CES), now available for everyone. 🔍 Why should you care? If you're building complex, intelligent systems with multiple LLMs or agents that must collaborate, delegate, and reason, ADK provides the full-stack tools to do it—all the way from development to deployment. 🧠 Core Pillars of ADK: 🔹 Multi-Agent by Design: Compose modular, hierarchical agents that delegate intelligently. 🔹 Model Flexibility: Use Gemini, Vertex AI Model Garden models, or any model via LiteLLM (Anthropic, Mistral, Meta, etc.). 🔹 Tooling Power: Use built-in tools (search, code execution), third-party tools (LangChain, LlamaIndex), or even other agents (LangGraph, CrewAI). 🔹 Multimodal Streaming: Built-in audio/video streaming for conversational agents. Just a few lines of code. 🔹 Rich Developer UX: Web UI, CLI, Python APIs. Visual step-by-step execution. 🔹 Built-in Evaluation: Define test cases and verify agent performance before deploying. 🔹 Easy Deployment: Containerize or deploy on Vertex AI’s managed runtime. 🧪 Example Use Case: A Weather Agent that Delegates With just a few lines of Python, you can build: ☞ A WeatherAgent that answers weather queries using a custom tool. ☞ A GreetingAgent and FarewellAgent that auto-handle greetings and goodbyes. ☞ All agents work together hierarchically, thanks to ADK’s intelligent delegation and orchestration. Here’s a peek: root_agent = Agent( name="weather_agent", model="gemini-2.0-flash-exp", tools=[get_weather], sub_agents=[greeting_agent, farewell_agent], description="Provide weather info, delegate greetings/farewells.", ) 🌐 Optimized for Google Cloud, Works Everywhere 📌 Deep integration with Vertex AI, Gemini, BigQuery, Apigee, and over 100+ connectors. 📌 Build once, deploy anywhere. 💡 Final Thoughts: Google’s ADK marks a pivotal shift from LLM-centric applications to multi-agent systems. This is ideal for those of us building modular, scalable, and intelligent AI workflows across diverse domains—from customer support to security monitoring and beyond. 🎯 If you’re serious about building agentic AI applications, now is the time to explore ADK. 📖 Dive in: Official ADK Docs - https://lnkd.in/gUVUBafE #AI #MultiAgent #LLM #GenAI #VertexAI #GoogleCloudNext #AgentDevelopmentKit #AutonomousAgents #OpenSource #Gemini #LangChain #MLOps #AIFramework #CTOInsights

110

29 Comments

Early in my career everything was about building large scale homogeneous systems. What’s been more interesting over the last 5y has been heterogeneous multi cloud systems. The tradeoffs are more complex to build simple (irreducible complexity) and elegant systems in that space.

41

2 Comments

With the Arm extension for GitHub Copilot, developers can access helpful guidance, optimizations, and code suggestions to make the switch to Arm architecture easier, all through GitHub. This extension is available on GitHub Marketplace and works with GitHub Actions along with the new Cobalt 100 native Arm Runners, creating a comprehensive CI/CD and MLOps pipeline. This means developers can enjoy a smooth workflow from edge to cloud, speeding up their time-to-market. Get set to embrace the future of software development on Arm! 👇

28