I joined Amazon in the US in September 2022, straight out of my master's program, and my reviews came back "Exceeds Expectations" two years running. A promotion I was expecting never came through, so I switched teams. My new senior manager looked over my promotion case, saw nothing missing, and intended to put me forward for promotion in March 2026. In January, though, the entire team was laid off, that manager too. My wedding in India was coming up in March, so losing the job right then piled financial and emotional stress on top of everything. And I hadn't done any coding, system design or behavioral practice since 2022.
For coding, the turning point was giving up on LeetCode as a heap of hundreds of separate problems. I zeroed in on the few patterns that sit behind most questions, learned how to recognize each of them, and solved a handful of typical problems for every pattern. After that, a new problem became something I could work through step by step, rather than something that only went well if I had seen it before. Two free video channels helped me a lot.
For system design, my main resource was an online course. Its biggest value to me was a repeatable game plan for the interview: pin down requirements, present a high-level design, hunt for bottlenecks, then go through deep dives and trade-offs. Thanks to that, I learned to take charge of the session instead of sitting back until the interviewer pointed me somewhere.
I began with the friendlier designs, Ticketmaster and Dropbox, only to get a feel for how an interview runs. When that felt natural, I moved on to the reusable design patterns and then came back to complete problems. By then the answers clicked, since I understood the building blocks and what interviewers want to see. I also subscribed to a system design newsletter as an extra, but the real game changer was writing out designs myself in the course's practice tool.
I also found other engineers who were interviewing and we took turns running system design mocks for each other. They gave me more confidence, better communication and better time management, and they surfaced weak spots I never would have caught on my own.
On the behavioral side, my years at Amazon paid off: I had authored design docs, led projects, partnered with other teams and seen how senior engineers get judged. Writing my own promotion document turned out to be hugely useful too. It forced me to revisit each project, separate my personal contribution from the team's, record how big and hard the problem was, explain the major decisions and trade-offs, and attach numbers to the outcomes.
Senior and mid-level hiring is mostly about whether you can take a fuzzy problem and break it down, decide without being told, work alongside or mentor other engineers, argue for your technical view, and ship without someone holding your hand. Starting from the promotion doc, I built a bank of stories tied to signals like ownership, delivering results, customer focus, ambiguity, disagree and commit, backbone, and running complex projects. Every story had the situation, my own actions, the trade-offs and a measurable result. In the interview, I worked out which signal a question was fishing for and picked the story that fit it best.
Visa's process opened with a chat with a recruiter, and then a HackerRank test lasting 90 minutes with two questions, a medium and a hard. The hard one was a graph problem, which I had practiced, so I had time to finish everything.
Roughly two weeks after passing, I got an invitation to a Visa Talent Day. The recruiter described it only loosely and said to expect a coding round plus low-level and high-level design. Friends at Visa had led me to expect a generic loop and then team matching. It went differently, because the interviewers were very interested in my recent agentic AI work.
Round 1 was coding plus a deep dive into my resume. The coding part was a medium question solved with two pointers and binary search, and it took me around 15 minutes. Then the interviewer spent a long stretch on an AI system listed on my resume. At Amazon I had built a RAG and agent system, end to end, to support on-call work; it took care of lower-priority tickets and replied to stakeholder questions in Slack. I had to sketch the architecture and explain it start to finish, and we covered prompt engineering, how we chose components and technologies, and the thinking behind the overall design. I had worked on it personally, so I could explain what we built and the reasons for it.
Round 2 was applied RAG implementation. The interviewer was one of the org's principal architects. He handed me some Python interfaces and wanted a RAG pipeline built, with pseudocode allowed. A portion of the code and a few components were provided, but the task mixed several technologies in a way unlike the internal Amazon tooling I was used to. At first I had trouble wiring everything together. I kept asking clarifying questions, took his hints on board and narrated my reasoning. I explained why embeddings are needed, how retrieval works, why cosine similarity can compare the query vector with the stored vectors, and when RAG is the right choice. The implementation wasn't familiar, but knowing the concepts and justifying each choice helped me recover and finish strong.
Round 3 was a system design session with the hiring manager, and it was among the toughest system design interviews I've done, counting those at other large tech companies. The prompt was to design a ride-sharing service like Uber. The framework from the course helped enormously. I led most of the discussion: I clarified requirements, grew the architecture a piece at a time, said what each component was for, went over trade-offs and flagged possible improvements. The hiring manager tossed in a few suggestions and I worked them into the design while keeping the thread of the whole thing.
What surprised me most was that this was not a plain coding and system design loop. Two of the rounds dug into the AI and RAG work on my resume, and the hiring manager was already in the mix, rather than showing up after some separate team-matching step. One day later, the recruiter got in touch to say the team wanted to give me an offer at the Senior Software Engineer level.
My advice starts with your resume: know each major project in depth. Be able to draw its architecture and discuss requirements, how data flows, how failures are handled, scale, security, observability, trade-offs and what alternatives you could have chosen.
Don't limit your prep to the categories the recruiter lists. Think of that outline as the floor, since interviewers can spend a long time on any specialized background you show on your resume.
Get the coding patterns down before you grind through piles of problems. With each one, practice naming the approach behind it, talking through your thinking, checking edge cases, and working out the time and space cost.
In system design, practice taking the lead. Drawing boxes and arrows isn't enough. The strong candidates clarify requirements, rank the important use cases, defend their choices, notice bottlenecks and nudge the interviewer toward deep dives worth having.
When you're stuck, stay talking and stay open. Ask pointed questions, think out loud, use any hints, and show you can take in new information. My second round started rocky, but explaining my reasoning and adjusting got me back on track.
Run mock interviews with other candidates. Understanding system design on paper and delivering a coherent design against the clock are separate skills.
For behavioral, assemble a story bank rather than drafting a different answer for each possible question. A compact collection of strong, versatile stories will stretch across ownership, conflicts, setbacks, ambiguity, leading others, customers and delivery.
Some wider thoughts. With so many layoffs, the market is now full of highly qualified people. Interviewing well might not be enough, because plenty of others can do equally well, and the outcome can hinge on relative performance, exact domain background, team fit, or things nobody on your side controls.
I've lived this. In an earlier process I was told I'd been picked, yet no offer followed, because the company turned up an internal candidate who already knew the product. I've also watched talented friends interview well and stay on the market for a long time.
Since so much is out of our hands, I prepare and apply side by side. Don't hold off on applications until you feel perfectly ready. Apply while you study, and treat each interview as practice for the next one. You don't need to win every interview, just one process that turns into the right offer.
Still, applying broadly shouldn't mean applying blindly. I noticed two styles of hiring.
One kind is hiring for a particular role or team. At companies like Apple, Oracle and some Uber teams, the hire can come down to a very particular skill set or product need. Resume fit counts for a lot there, so apply when your background truly lines up and tailor the resume to make that obvious. I also went for several fitting openings at those companies, where separate teams seemed to assess candidates independently instead of restricting you to a single active role. For Apple and Oracle, at least at the moment, there's no cooldown period.
The other is general, company-level hiring. At places like Amazon and Google, the first evaluation seems to rest more on typical coding, design and behavioral bars, either ahead of or together with team matching. A strong resume won't carry you there. You have to know what the loop is measuring and prepare to demonstrate exactly those signals.
So I'd say apply a lot, but with aim. Keep applying to every role that really fits your background, learn how each company hires, and prepare both for the topics on your resume and for what that company typically asks.
One last thing: a rejection doesn't mean you interviewed badly or aren't qualified. Another candidate may simply match the product more closely or come from inside. Keep getting better, keep sending applications, and keep widening your pipeline until one of them lands.
Discussion
Loading comments…