#Top 200 questions and company tags

Read the source note first. It changes how much weight to put on the company column.

#1 · Where this data comes from — and where it does not

This is not Big Omega data, and it is not LeetCode Premium frequency data. Those datasets are paid, scraped, and not something this handbook can read or republish accurately.

What this list actually is:

Treat the company column as directional, not authoritative. Company question banks rotate, leak, get retired, and vary by team and by office. A tag here means this problem has been widely reported at that company, not this company will ask you this. If you want live frequency data, LeetCode Premium's company filter is the real source — this list is for deciding what to practise, not for predicting your interview.

The useful signal is not the company. It is the pattern. Companies do not share a question list; they share a pattern distribution. Every problem below is tagged with the pattern it drills, and that is the column to plan from.

#2 · Coverage

The handbook's pattern pages reference 258 distinct LeetCode problems. Against this 206-problem list:

The last row is not padding. The pattern pages include ladder rungs and teaching problems that are not interview-frequent but build the intuition the frequent ones need.

#The gaps

The problems below appear on the top list but are not yet worked into a pattern page. Each is still listed in the tables further down, with the page whose pattern it belongs to — so you can slot it into that page's ladder.

None of the gaps is a missing pattern. They are additional problems in patterns the handbook already teaches — mostly matrix manipulation, extra grid-DP variants, and tree traversals. If you can do the worked examples on the relevant page, these are reps rather than new material.

#3 · The list, by pattern

✓ = taught on a handbook pattern page.

#Arrays & hashing — [hashing.html](hashing.html)

#Array manipulation — [hashing.html](hashing.html)

#Strings — [strings.html](strings.html)

#Two pointers — [two-pointers.html](two-pointers.html)

#Sliding window — [sliding-window.html](sliding-window.html)

#Prefix sums — [prefix-sum.html](prefix-sum.html)

#Stack & monotonic stack — [stack.html](stack.html)

#Binary search — [binary-search.html](binary-search.html)

#Linked lists — [linked-lists.html](linked-lists.html)

#Trees — [trees.html](trees.html)

#Tries — [tries.html](tries.html)

#Heap & top-k — [heap.html](heap.html)

#Intervals — [intervals.html](intervals.html)

#Backtracking — [backtracking.html](backtracking.html)

#Graphs — [graphs.html](graphs.html)

#Union-Find — [union-find.html](union-find.html)

#Greedy — [greedy.html](greedy.html)

#Dynamic programming — [dynamic-programming.html](dynamic-programming.html)

#Bit manipulation — [bit-manipulation.html](bit-manipulation.html)

#4 · By company

Same caveat as above — directional, not a prediction.

Read the pattern distribution, not the problem numbers. The tag data is coarse: Amazon interviews broadly and is reported on almost everything here, so its row carries little information. What survives that noise is the relative pattern weighting, and that is the only column worth planning from.

What to take from this table: every company's top patterns are drawn from the same small set — trees, graphs, DP, hashing, two pointers. That is the actual finding, and it is why the handbook is organised by pattern rather than by company. There is no company-specific curriculum to learn.

What the table does support: Google and Amazon lean hardest on dynamic programming and graphs; Meta, Microsoft and Bloomberg lean on trees and linked lists. If you are short on time and interviewing at one of the first two, weight DP and graphs; at the second three, weight trees and pointer work.

What it does not support: anything about the rows with fewer than about 40 tags. Uber at 26% backtracking is 4 problems out of 15 — that is sample size, not a hiring signal. Read only the top five rows as meaningful.

#5 · How to use this

Do not work through it top to bottom. 200 problems attempted once is worth less than 60 problems you can re-derive, and the whole practice method is built on that claim.

If you have a named company: read its row in the table above for the pattern weighting, and drill the two or three heaviest. The specific problem numbers tell you almost nothing — that list will have rotated by the time you sit down, and the pattern distribution will not have.

The order within a pattern matters more than the order between patterns. Each pattern page's ladder is built so that each rung teaches something the next one assumes. This page is an index; the ladders are the curriculum.