LeetCode
A structured approach to algorithmic problem solving.
Pat's LeetCode Status: 125
150
300
500
1000
Quality over quantity
It's much better to deeply understand 300 problems than to do 1000 shallowly.
Be a master of 150, not a silver of a 1000.
Remember this.
Timers
| Difficulty | New Problem | Revisit |
|---|---|---|
| Easy | 15 min | 5 min |
| Medium | 25 min | 10 min |
| Hard | 40 min | 20 min |
Beginner Tips
From a Google Engineer:
- Take it easy - it's okay to feel overwhelmed at the start. Relax, take a sip of tea, and do your best.
- If an algorithm is named after someone, you can safely skip it.
- Name all variables descriptively.
- When doing NeetCode 150, think about how the section's previous exercises apply to the current problem.
- Struggle productively and be consistent.
What's neetcode?
Think of it as repository of solutions to LeetCode exercises.
Highly recommended for their roadmap which gives structure.
The Process
1. Understand the Problem
- Read the problem statement twice
- Summarize in two sentences: what's the input, what's the required output?
- Work on paper a small example
- Write the constraints
2. Come Up With a Solid Plan
- What do you know about the problem?
- What's the one small thing you can try?
- First, solve it in any way - brute force, then optimize
- The logic must be sound before implementation begins
3. Only Then, Write Code
- Debugging should be minimal if the plan is solid
- Never randomly change code (e.g.,
-1to+1,<=to<)
4. If Not Solved in Time
- Review the solution and find the delta between your approach and the answer
- Take note of the delta
- Create an Anki card with the problem and delta
About LeetCode Metrics
LeetCode's time & memory measurements are notoriously noisy and unreliable. The percentiles are not useful for interview prep or understanding algorithmic efficiency. In fact they can be harmful.
Constraint to Complexity Cheat Sheet
| Input Size (n) | Max Viable Complexity | Common Approaches |
|---|---|---|
| n <= 10 | O(n!) | Brute force, permutations |
| n <= 20 | O(2^n) | Backtracking, bitmask DP |
| n <= 100 | O(n^3) | Floyd-Warshall, triple nested loops |
| n <= 1,000 | O(n^2) | Nested loops, simple DP |
| n <= 10,000 | O(n^2) borderline | Optimized quadratic, may need O(n log n) |
| n <= 100,000 (10^5) | O(n log n) | Sorting, heaps, divide & conquer |
| n <= 1,000,000 (10^6) | O(n) | Linear scan, hash maps, two pointers |
| n <= 10^7+ | O(n) or O(log n) | Math tricks, binary search on answer |
Quick Heuristics
- 10^8 operations ~ 1 second (rough ceiling for most judges)
- If n^2 > 10^8, you need a better approach
- Sorting is "free" at O(n log n) - often a good first step
- Hash map gives O(1) lookup - turns O(n^2) into O(n)
- If stuck, ask: "What would I need to make this O(n)?"
Interview Specific Drills
- Set a timer to beep every 60 seconds to build the habit of talking while coding
- Every beep you need to summarize your current state in one or two sentences
LLM Review Prompt
Use this prompt to get structured feedback on your solutions:
Click to expand prompt
You are a senior engineer conducting a coding interview at a top-tier
company (Google, Anthropic, Jane Street caliber). Review the provided
solution against elite hiring bar standards.
If the problem statement is not provided or cannot be clearly inferred
from the code, ask for it before reviewing.
## Review Structure
**Score (0-100)**
Anchors:
- 90+: Strong hire signal, demonstrates senior-level thinking
- 75-89: Would advance, minor concerns
- 60-74: Borderline, notable weaknesses but fundamentally sound
- Below 60: Would not advance
**Analysis**
1. Correctness: Edge cases, off-by-ones, boundary conditions
2. Complexity: Time/space analysis. Flag if suboptimal approach exists.
3. Code quality: Variable naming for production (not single letters
except loop indices), structure, readability. Interview context -
no need for docstrings, but logic should be self-evident.
4. Missing elements: Imports, type hints if partially used, inconsistencies
**Weak spots** (for targeted practice)
Bullet specific patterns or concepts to reinforce. Keep actionable.
**Verdict**
If the solution is correct but wouldn't impress, say so explicitly and why.
**Optional follow-up**
One likely interviewer probe (e.g., "How would this behave with 10^8
elements?" or "Can you do this in-place?").
---
Context: This is part of an elite engineering training program targeting
1850+ LeetCode elo and senior SRE/MLE roles. Optimize feedback for
learning, not ego.
Problem Lists
Looking for structured paths? Check out LeetCode Paths in Resources for curated problem lists like NeetCode 150, Blind 75, and more.