The complete interview preparation guide

Almost every product company runs a version of the same funnel. Knowing the stages lets you prepare the right thing at the right time instead of grinding LeetCode and hoping it covers everything.

The stages are not weighted equally, and the weighting shifts with seniority. For a fresher or SDE-1, coding rounds dominate and a weak system design round is survivable. From SDE-2 upward the design round carries as much weight as coding, and at staff level it usually decides the outcome. Behavioral is a genuine gate everywhere, not a formality — more on that below.

Candidates consistently underestimate the calendar. From first contact to a signed offer, budget four to ten weeks per company. A recruiter screen leads to an online assessment within a week or so; the onsite loop lands two to three weeks after that; the debrief and hiring committee take another one to two weeks; negotiation adds a few days more.

Then add the Indian notice period, commonly sixty to ninety days. If you want to be in a new seat by a particular month, work backwards: subtract the notice period, subtract another two months for the interview pipeline, and subtract your preparation window. That is when you should have started. Most people begin far too late and end up interviewing underprepared because a recruiter happened to reach out.

Interview in parallel, not in series. Offers that arrive within the same two-week window give you real negotiating leverage; offers spread across three months give you none, because the first one expires before the second arrives.

If you have a full-time job, you realistically get about two hours on weekdays and four to five on weekends — call it fifteen to twenty hours a week. Do not plan as if you have unlimited time. Plans built on four weekday hours collapse by week three, and the collapse itself is demoralising enough that people quit the whole effort.

Allocate the bulk to consistent DSA pattern practice, a steady weekly block to system design fundamentals, and a smaller recurring slot to behavioral story-building. The behavioral slot is the one everybody skips, and it is the cheapest marginal improvement available — a few hours of story preparation moves a whole round.

Consistency beats intensity. Five focused days a week for twelve weeks comfortably beats a heroic two-week sprint, both for retention and for the calm you carry into the room. Retrieval practice — attempting a problem cold before looking at the solution — is what actually builds recall. Reading solutions feels productive and teaches you very little.

If you need a default structure, this one works for a working engineer targeting a mid-to-senior product-company role. Adjust the weighting to your gaps rather than following it blindly.

The hardest filter in the whole pipeline is the resume screen, and the most reliable way through it is a referral. A referred application is read by a human; a cold application often is not. This is the single highest-leverage hour in your entire preparation, and most engineers spend zero time on it.

You do not need a close friend at the company. A polite message to an engineer in a relevant team, with a specific reason you are interested and a one-line summary of what you have built, converts far better than people expect. Make it easy to say yes: attach the resume, name the exact requisition, and accept a no gracefully.

Keep a target list of fifteen to twenty-five companies rather than applying to everything. Tiered by preference, with the weakest tier interviewed first so your worst performance happens where it costs least.

Recruiters spend seconds per resume, and an ATS may parse it before any human sees it. Lead every bullet with impact and a number, name the technologies explicitly, and keep it to one page unless you are staff-plus.

The formula that works is: accomplished X, measured by Y, by doing Z. "Reduced p99 API latency 40 percent by introducing a Redis cache layer and batching downstream calls" beats "Worked on backend performance" in every dimension — it is specific, it is quantified, and it names the technology an ATS is scanning for.

The OA is a filter, not an evaluation. It is typically sixty to ninety minutes for two problems, auto-graded against hidden test cases, and often scored on partial credit for cases passed rather than all-or-nothing.

Two practical consequences. First, practise under an actual timer, because time pressure is the thing being tested and it changes how you think. Second, get a correct brute-force solution passing before you attempt to optimise — partial credit for a working slow solution beats zero for an elegant one you did not finish. Read all problems before starting and do the easier one first.

Interviewers are not scoring whether you produced the optimal solution. They score along several axes at once, and a candidate who solves the problem in silence frequently loses to one who solves a slightly weaker version while communicating clearly.

When two candidates are technically comparable — which happens constantly at the offer boundary — the behavioral feedback decides. At Amazon the leadership principles are weighted at least as heavily as the coding rounds and a bar raiser can veto on them alone. Most engineers walk in having prepared nothing.

The fix is cheap: five to seven real, quantified stories from your own work, each mapped to several themes. That is a few hours of work for a round that carries as much weight as the ones you spent twelve weeks preparing for.

Small operational failures cost real points, mostly by rattling you in the first ten minutes when you are already tense. Handle them the day before.

The Indian offer structure has more moving parts than base salary, and each is a separate lever. Total CTC quoted by a recruiter often bundles base, a variable or performance component, a joining bonus, a retention bonus, and equity — and those are not equally valuable. Variable pay is conditional. A joining bonus is one-time and often clawed back if you leave inside a year. Equity depends entirely on the vesting schedule and whether the stock is liquid.

Ask for the full written breakup before responding to anything. Compare offers on the components that actually recur — base plus reliably-paid variable plus annualised equity — not on the headline number, which is the one designed to look largest.

A competing offer is by far your strongest leverage, which is the real reason to run your interviews in parallel. Notice period is negotiable too and is worth money to the hiring side: a shorter buyout is something they can often fund. Never accept on the call. Thank them, ask for it in writing, and take a day. Nobody has ever lost an offer by asking for twenty-four hours.

You will be. Strong engineers get rejected by companies that later hire them. A rejection is one loop, on one day, with a particular set of interviewers and a particular problem, and the noise in that process is much larger than candidates assume.

Ask the recruiter for feedback — sometimes you get something specific and useful. Note the cooldown period, typically six to twelve months, and diarise it. Then write your own debrief while it is fresh: which round felt weak, which question you fumbled, what you would do differently. That log is worth more than the recruiter feedback, and after three or four loops the pattern in it tells you exactly what to work on.