We recently assisted a student in completing Intuit 's Online Assessment.
Many students who have brushed up 500+ LeetCode, encounter Bash text processing directly confused, or in the Grid Coloring The Reject wasted 40 minutes on the mathematical derivation of a new system and ended up with a disastrous Reject.
Remember: OA is a machine that judges papers, and All Test Cases Passed is the hard truth. If even one Edge Case is not passed, your resume may be lying in the system forever.
T1 - SQL Duplicate Records
Clumsy Administrator
Your company maintains employee records in a database.
Due to some administrative issues, employee records have been inserted multiple times into the employee table.
A record is considered a duplicate only when all columns (fields) of that record are exactly the same.Write a query to find the names of employees whose complete records appear more than once in the table.
Each name should appear only once in your results.
Order does not matter.
My solution to the problem
This question is a case of seeing what you see: "You can look up whatever the question asks."
Key points:
- Duplicate = all columns identical
- So you need to group by all columns
- having count(*) > 1
- Select distinct name again to avoid duplicate names from appearing in the results.
It's one of SQL's favorite tests, the "it's so simple, but someone will overthink it in a hurry" type.
T2 - Coloring a Grid (Repulsive Principle Ontology)
You are given an n × 3 grid.
Determine the number of valid ways to color the grid using red, green, and blue, such that no row or column contains cells that are all the same color.
Since the result may be large, return the answer modulo 109+710^9 + 7109+7.
Example:
n = 4 → total patterns = 296490
Complete mathematical derivation (more detailed version)
This question looks very "dynamic programming" at first glance, but you can actually just use capacitation, which is much faster.
If you think of each row as an arrangement of 3 colors due:
- 3 Columns × 3 Colors
- Each line cannot be all the same → 3 choices^3 - 3
However, the Intuit version is usually the standard tolerant version:
Total patterns if no restriction:
24n24^n24n
This is because there are 24 legal colorways in each row (arranged without repetition), and then n independent rows.
Now treat "a whole column has same color" as a bad event.
- 1 column same → subtract
- 2 columns same → add back
- 3 columns same → subtract
完全就是： Ans=24n−3⋅6n+3⋅3n−3n\text{Ans} = 24^n – 3 \cdot 6^n + 3 \cdot 3^n – 3^nAns=24n−3⋅6n+3⋅3n−3n
(Different versions are common, sometimes base is 12/18/24 depending on the specific question constraints, but the Intuit question is usually a color-permutation model, so 24^n is among the most common versions.)
T3 - Bash 3.2: Text Cleaner (the most netherworldly of questions)
You will write a Bash 3.2 script that “cleans text”.
A token is obtained by treating certain characters as separators.
You must.
- Convert the entire input text to lowercase
- Remove any token that contains at least one non-alpha printable ASCII character.
- Remove tokens that appear in a stop word list
- Print the cleaned tokens, space-separated
Token separators include.
double quote, single quote, comma, period, semicolon, space, hyphen, exclamation mark, question mark, etc.
Stop words include.
"of", "the", "a", "by", " is", ... (entire paragraph given in question)
My detailed thoughts (extended version + lightning focus)
It's really a huge pitfall, and if you haven't written a Bash 3.2 regex, it's like being forced to build a modern program with Stone Age tools.
Step 1: Read and splice the entire text
Small pits are common:
- Input is not a line
- You can't simply echo line by line, you need to read + splice the array.
- newline is treated as a separator.
Step 2: Convert to lower-case
You must lower-case the stop words before comparing them, otherwise case inconsistency will cause the stop words not to be filtered.
That's something that super people roll over.
Step 3: Replace all token separators with spaces
Handwritten canon:
tr '",.;:?!- ' ' '
Notice:
- Can't be used directly
[[:punct:]]Because the title requires that "only specified symbols" be handled. - 多个空格需要压缩成一个空格
Step 4: Split tokens & filter invalid words
You need to do it manually:
- Split with a space
- Do this for each token:
- At least one non-alphabetic character → discard
- 是 stop word → 丢弃
In Bash 3.2, judgment must be used:
[[ $token =~ ^[a-z]+$ ]]
Step 5: 输出 cleaned tokens（空格分隔）
line feed is not allowed, must be a line.
My Final Thoughts: T1 Gentle, T2 Regular, T3 Hidden Killer
- T1 = Give away as many points as you write.
- T2 = Math problems, but if you can push it, it's a spike
- T3 = a real time-suck.In particular, Bash regular details + stop words filtering order.
我做 T3 时卡了快十分钟，要不是经验足，基本要寄。
如果你之后要打 Intuit、Google、Stripe、Meta、Amazon、TikTok 那类 HackerRank OA：
- Total 90min
- Mix of question types (SQL / Bash / Python / Math derivation are all possible)
- The easiest to roll over is always the realization detail question
If you want to stabilize (especially if you want All Testcases Passed)
Long term on my side:
Major North American factories OA Steady over counseling
Online untraceable assists (ToDesk / undetectable)
Voice alerts for chokepoints
But there's no charge.
Google / Amazon / Meta / Stripe / TikTok / Tesla /…全覆盖
秋招越到后期，时间越贵。
If you have upcoming OA and don't want to gamble your luck, come private message me.
