|
Hello Reader, I have interviewed over 300 solutions architects at AWS. Most of them knew more AWS than I did. Most of them did not get the job. One candidate drew me a perfect multi-region architecture in four minutes. I still said no. The next one drew something messier, and I fought to get her hired. Nothing about the second diagram was better. She did a few things in that room that most candidates never do. Here is what almost nobody understands about the interview room. The whole time you are talking, I am writing notes. Almost none of those notes are about how many services you can name. I am not writing down "mentioned Kinesis, mentioned Bedrock, got VPC peering right." I am writing down the things you actually did in the room. That is the whole game. Here are the five habits the top 1% have, and where each one shows up in the interview. None of them require you to learn a new service. Habit one: ask before you drawThis one happens before the candidate draws a single box, and it is where most people lose the system design round in the first two minutes. When the top 1% get a design question, they spend the first couple of minutes not designing anything. They ask how many users. What region. Is there a compliance requirement. What already exists. Who owns what. Only then do they go to the whiteboard. Most people, when asked to design an e-commerce website, start drawing straight away. And it does not matter how good the drawing is. Once they do start drawing, every service comes with a why. You can build any architecture with five equivalent services. What I want to see is whether you know the trade-offs and when to pick what, based on the functional and non-functional requirements. So do not say "I will use DynamoDB." Say "I will use DynamoDB here because we have a thousand different product types, they are read constantly, no joins needed. DynamoDB gives me single-digit millisecond reads and scales to millions of users on its own." Then the non-functional layer. Security is our first priority, so I will lock DynamoDB down with IAM and a VPC endpoint, and I will draw that in. That is how a real architect works on a real project. You are just showing the same thing in the room. That is how you get hired. Ask first, then give a why for everything you draw. The diagram comes last. Habit two: a bigger diagram is a worse diagramA lot of candidates think complexity signals competence. The top 1% go the other way. Amazon has a leadership principle called invent and simplify, and every top company has a version of it. If you walk in with a really complex architecture and a lot of services, that is not a point in your favor. Anybody can build something complex. A good architect gets the same result with fewer components, because in the real world every service you add is more overhead and one more thing that can fail in production. Here is the pattern I see constantly. I ask somebody to design a microservice and they bring in Kubernetes, or a queue, or a stream, and the use case does not need any of it. Then I ask how they would scale that Kubernetes workload, how they would secure it, how they would cost optimize it. And there is nothing there. They just wanted me to hear the word Kubernetes. We do not reject you for saying load balancer and EC2. We reject you for saying Kubernetes and not being able to defend it. If you have ever wondered why your coworker got into big tech knowing fewer services than you, this is why. And if you do overdesign, a good interviewer will usually give you a chance to fix it. Watch what happened with one candidate. He drew load balancer, EC2, Kinesis, Lambda, RDS, DynamoDB, EventBridge Pipes, S3, EMR. I asked him, do you need all of this? He paused and said: "Actually, this is a shopping cart. I don't need streaming. Let me take out Kinesis and the Lambda behind it. ALB, EC2, RDS does it synchronously. I was putting data in S3 for analytics, but if nobody's asking for that insight, that goes away too. Is that what you were pointing at?" That is a great answer. He took the hint and simplified it against the requirement, and then he ended on a question back to me, which turns a quiz into a conversation. When you can turn the round into a conversation, your chances go way up. So do not add services just to sound smart. Use the fewest you need and be ready to tell me why each one is there. Habit three: answer at the right altitudeThe first two habits are about what you draw. This one is about who you are explaining it to. In an interview process, you will talk about the same big project with a lot of different people. Most candidates explain it the exact same way, no matter who is in front of them. That is a mistake, because the recruiter is listening for one thing, the hiring manager is listening for something else, and the technical interviewer wants something completely different again. Most candidates only have the technical version. They are out before they even reach the technical round. This was the hardest thing for me when I switched into architecture. I came from a deep technical background, and I used to think people who could not code had no business being managers or directors. That is just wrong. Every role has its place. Technical people deliver the application, but without the business side there is no customer and there is no money coming in. Think about the actual job. You spend half your week with senior leaders working out what the organization wants. Take the same project, moving 100 microservices to either EC2 or Lambda, and watch how it changes by audience. To the recruiter: "In my most recent project I had to choose between EC2 and Lambda to modernize 100 microservices. I connected the customer experience, the business need, and the technical trade-offs, did a proof of concept, got buy-in from the execs and the engineering teams, and we went with Lambda, API Gateway, DynamoDB, CloudFormation and CloudWatch alongside it. There are public artifacts from this project on my GitHub and LinkedIn." Notice what that did. Business context, executive buy-in, technically competent, and a link to proof, because recruiters look for the GitHub. No technical mumbo jumbo. Same opener to the technical interviewer, then: "EC2 is a virtual machine. I can install anything, and if the workload is already on a virtual machine, the migration is easier. Lambda is serverless, no servers to manage, but there are limits like the 15 minute maximum runtime, and I have to think about cold starts and concurrency." Then the design factors, then why Lambda won. Same opener to the hiring manager, then: "I treated it as a business and engineering decision. I set the modernization goals and constraints, compared both services against them, looked at the migration effort and the operating cost. I delivered it on Lambda as a hands-on architect, documented the learnings, trained the team on them, shipped in three months, and our vice president recognized it." If you cannot explain your architecture to someone who cannot code, you do not own it yet. There are a couple more best practices and more details on each. If you are walking into an interview in the next few weeks, watch it before you practice anything else: Keep learning and keep rocking π, Raj P.S. If you have found this newsletter helpful, and want to dive deeper... Everything I teach on AWS, System Design, Kubernetes, and DevOps - all in one Skool classroom. Build the technical depth at your own pace, and get every new course I release: Join Hereβ π₯ Free tutorials on Cloud Gen AI + interview prep on my YouTube channel: Hereβ π The SA Bootcamp is the full transformation - hands-on practice through Cloud Colosseum, interview prep, mock interviews, and resume reviews, and so much more... βJoin the waitlist to be first in when the next cohort opens. |
Free Cloud Interview Guide to crush your next interview. Plus, real-world answers for cloud interviews, and system design from a top AWS Solutions Architect.
Hello Reader, Two weeks ago I was at the AI Engineer Conference in New York. I gave a talk on agent memory and building a second brain for AI agents, with a live demo. The talk was fun. The hallway conversations were even better. I spent most of my time talking with Forward Deployed Engineers and engineering leaders from companies that are running AI in production today, and a clear picture came out of those conversations. In todayβs newsletter, I will share my biggest takeaways and what they...
Hello Reader, I ran GPT-6 Astra through the exact Solutions Architect interview I used at AWS. Same five rounds, same rubric, same interruptions I give real candidates when they ramble. I wanted to know something specific: not whether it knows AWS, but whether it would pass the bar that actually gets someone an offer. It passed some rounds cleanly. It failed the ones that matter most. Here is exactly where, and why the gap is not something a smarter model fixes. Round one: tell me about...
Hello Reader, I'm growing the team at Cloud With Raj and I'm looking for someone to teach live classes inside our AWS Solutions Architect Bootcamp. This is the program where our students go from stuck in IT to landing $200K+ cloud roles at companies like AWS, JPMorgan, Siemens, Capital One. And you'd be one of the people helping them get there. π§ What you'd actually be doing You'd be hosting 2-3 live classes per week covering AWS architecture, system design, GenAI on AWS (Bedrock, RAG,...