Cracking the Coding Interview

Preparing for technical interviews

1. The Interview Process ::Pg.4::

1.1 What

Most top tech companies and the many other companies that copy them use algorithm and coding problems to as the largest component of the interview process.

These boil down to problem solving questions testing the candidates ability to solve algorithmic problems you haven't seen before. This aims to assess based on the following categories:

Analytical Skills:
- Optimal solution
- How much help/hints were needed
- How long it took
- etc.
Coding Skills:
- Translating algorithm to code
- Considering Tradeoffs
- Followup to design/architect a new solution
- etc
Technical Knowledge / Computer Science Fundamentals:
- Strong fundamentals
- etc
Experience:
- Good technical decisions in past
- Built interesting
- Challenging projects
- etc
Culture fit / Communication skills:
- Personality and values
- Fit with the company
- etc

1.2 Why?

Because its better than nothing. Companies need some way to filter the thousands of candidates per position. No point in complaining, just study and prepare for the interviews.

1.3 How Questions are Selected

Some companies have their own internal question banks, some use public question banks, and some use a combination of both.

Prepare for both cases:

Current up to date company question
Popular and important leetcode questions

1.4 Its All Relative

For a specific question, an interviewer is simply comparing you to other candidates:

candidate 1: 30 mins to solve
candidate 2: 50 mins to solve
candidate 3: 15 mins to solve but many hints
candidate 4: 10 mins with an alternate approach

A hard question thus isn't a bad thing. When its harder for you, its harder for everyone else as well. An interviewer is simply getting a feel for you performance relative to all other candidates they have ever asked this question to.

1.5 Frequently Asked Questions

No answer, rejection? If you haven't heard in 3-5 business days, check in with the recruiter, else move on.

Reapplying? Yes, but typically have to wait 6 months to a year.

2. Behind the Scenes ::Pg.8::

2.1 A Repeat Curtain

Most companies conduct their interviews in similar wars:

Round 1. 2x Screening Interviews

coding and algorithms questions
bar can be just as high as onsite interviews

Round 2. Onsite 4x-6x Interviews

the standard technical interview

Round 3. Lunch Interview

non technical
discuss interests
opportunity to discuss company culture

Round 4. After Interview Feedback

performance
feedback
decision

2.2 Per Company Basis

Excluded this for now

3. Situations

3.1 Experienced Candidates

More experienced engineers might see less algorithm focused questions or be held to a lower standard as its been a few years since they took an algorithms class.

Or on the other hand, they might be held to a higher standard as more years of experience allow a candidate to see more problems.

Either way, it balances out.

System design is the exception where its more for experienced engineers since design is something that is usually not covered in university.

If a question is too easy, candidate performance gets clustered together

Look For Questions With Multiple Hurdles:

Some questions have a single "Aha!" moment which require a specific insight
A series of hurdles, insights, or optimizations are better than a single insight
If a single hint makes a substantial difference in a candidates performance, then its usually not a good question

Use Hard Questions, Not Hard Knowledge:

Most cs graduates understand trees but may not remember dijkstras algorithm
Check if knowledge you are testing is an important skill

Avoid "Scary" Questions:

Some types of questions intimidate candidates because it seems they involve specialized knowledge (even though usually they don't)
Math or probability, low-level knowledge, system design or scalability, etc

Offer Positive Reinforcement

Many candidates are intimidated by interviewing and try to read into the interviewer's every word.
You want a candidate to feel comfortable as a nervous candidate will perform poorly
Try to be warm and friendly to your candidate

Probe Deeper On Behavioral Questions

Many candidates are poor at articulating their specific accomplishments
Usually you ask them about a challenging situation, and they tell you about a difficult situation their team faced, and as far as your can tell they didn't really do much.
Not describing the details of the actions they took make them a flawed candidate, but not necessarily a flawed employee.

Coach Your Candidates

Many candidates don't use a good example to solve an interview questions
Some candidates take a long time to find a bug because they use an enormous example
If they dive into code before finishing planning the optimal solution, pull them back and have them focus on the algorithm
If they get nervous and stuck, suggest that they walk through the brute force solution, as their first solution doesn't have to be perfect

If They Want Silence, Give Them Silence

Learn to distinguish between "I'm stuck" and "I'm thinking in silence"

4 Types Of Questions

Sanity Check

Easy problem solving or design questions.
Assess minimum degree of competence to filter out the worst candidates

Quality Check

More challenging, again problem solving or design
Make a candidate think

Specialist Questions

Test specific tops such as Java or machine learning
Used for skills even a good engineer can't learn quickly on the job

Proxy Knowledge

Knowledge that is not quite at the specialist level
But if a candidate has worked in depth with these technologies and can't tell you why a specific choice is good or bad, that suggests and issue that they are not absorbing information core to their job

Textbook: Cracking the Coding Interview