A twelve-week technical interview plan that respects your time

Spend the first three weeks on arrays, strings, hash maps and two pointers. That is it. These four topics account for the majority of screening-round questions across Indian product companies and service companies alike, and comfort with them removes most of the panic from a live round.

Work in a loop of solve, then re-solve. Attempt a problem for twenty-five minutes without help. If you are stuck, read the approach, close it, and implement from scratch. Then, and this is the part almost everyone skips, put it in a list and solve it again from blank three days later. The second solve is where learning actually happens; the first is exposure.

Aim for around forty problems in this phase, each solved twice. Forty problems understood beats two hundred and forty copied.

• Weeks 1 to 3: arrays, strings, hash maps, two pointers.
• Twenty-five minute attempt, then read, then re-implement.
• Every problem gets a second solve three days later.

Now add linked lists, stacks and queues, binary trees and basic recursion, including binary search on both arrays and answers. Recursion is where most candidates lose a round, not because it is conceptually hard but because they have never had to say the recurrence out loud while typing.

Practise narrating. Set a timer, open a blank editor, and explain your approach aloud before writing anything, then keep talking as you write. It feels ridiculous alone in a room and it is the single highest-return habit in interview preparation, because interviewers score communication as much as correctness.

Add one weekly mock: a friend, a peer group, or a recorded self-session with a random problem you have not seen. The purpose is not to solve it, it is to practise being observed.

Graphs reduce to a small set of patterns, breadth-first traversal, depth-first traversal, cycle detection, shortest path on unweighted and weighted edges, and union find. Learn the patterns rather than individual problems and you will recognise most variants.

Dynamic programming deserves less time than its reputation suggests for most roles below senior level. Cover the classic ladder, climbing stairs, coin change, longest common subsequence, knapsack, and the two-dimensional grid family. Ten well-understood DP problems is a reasonable target; chasing fifty is a poor use of three weeks.

In parallel, prepare your projects properly. For each project write down the problem, your specific contribution, one hard decision with the alternative you rejected, one thing that broke in production or testing, and what you would change now. Interviewers dig into projects far more than candidates prepare for, and 'I would refactor it' is not an answer.

• Graph patterns: BFS, DFS, cycles, shortest path, union find.
• DP: ten canonical problems, understood, not memorised.
• One page of notes per project, including a failure you fixed.

Add a light system design layer, even for entry-level roles, because the questions now show up early. You need a working vocabulary: client and server, load balancer, caching, database indexes, read replicas, queues for asynchronous work, and the trade-off between consistency and availability. You are not expected to design a global system; you are expected not to be blank.

Behavioural rounds in India are frequently underestimated. Prepare six stories, structured as situation, task, action, result: a conflict, a failure, a deadline you saved, something you learned quickly, something you taught, and a decision you made with incomplete information. The same six stories answer almost every question asked.

The final two weeks are rehearsal, not new material. Two full mock interviews a week, a review of your problem list, and a re-solve of everything you failed twice. Learning new topics in the last fortnight generally increases anxiety more than capability.

Advanced segment trees, heavy number theory, obscure string algorithms and competitive programming tricks are not on the path unless you are specifically targeting a competitive-programming-driven employer. They consume time that is better spent making your fundamentals unshakeable.

Skip the accumulation of paid courses. The marginal value of a fourth course is near zero compared with a second pass over problems you already failed. Preparation is limited by revisiting, not by acquiring.

Skip comparing your problem count with anyone else's. Counts are a vanity metric and they correlate poorly with performance in a live round.

Write a post-mortem within an hour of every interview, while the questions are fresh: what was asked, where you hesitated, what you would say differently. This document becomes your real syllabus and it is far more targeted than any generic list.

Expect a low conversion rate and plan volume accordingly. Even strong candidates convert a minority of interviews, and a rejection frequently reflects a headcount decision rather than your performance. The plan works when it is repeated, not when it succeeds first time.