[Samurai]

This time I resisted asking the 20th century question for a month. Finally, I asked it today to a FCD candidate. [FCD = First Class with Distinction]

Q: Which technology had the greatest impact on 20th century? And how?
A: Smartphone
Q: Um, when was 20th century?
A: 1980-1990
Q: What is the current century?
A: 22nd century.
Q: If 1980-90 was 20th, how can this be 22nd? Please work it out the white board.

The candidate found the arithmetic mistake once 3 centuries were written down, and changed 20th to 1990-2000.



And we continued....

Q: Do you think smartphones had the maximum impact on 1990-2000?
A: Um, don't know. May be 22nd century.
Q: Alright, we'll settle for that.

Finally, I didn't hire the candidate because of profile mismatch. I need people who are observant/analytical enough to know century has 100 years. However, service companies like TCS/Infy can make good use of such candidates where the work is well defined and no analytical skill is required.

[Samurai]

Originally Posted by Samurai

Me: You have consistently got 90+ in mathematics every year.
She: Yes, sir.
Me: You know Fourier Transforms well?
She: Yes sir.
Me: What is Fourier Transforms used for?
She: Hmm, don't know.

Originally Posted by ampere

Fourier and Laplace Tranforms form the core of electrical engineering concepts for signal processing and circuits. An electrical/electronics engineering candidate simply cannot say he does not know these concepts.

This time I came across another candidate with insane score in mathematics in every semester. Since she was an electronics and communications engineering graduate, I had to ask...

Me: Do you know Fourier Transforms?
She: No
Me: But, your scores in mathematics is too good, how did you not learn Fourier Transforms.
She: Our maths lecturer asked us to skip it in choice. Even without studying FT, we could aim for 100 percent marks.
Me:

So... we have maths lecturers in engineering colleges telling electronics and communications engineering students not to bother learning Fourier transforms. I am sure that idiot of a lecturer didn't know that Fourier transforms forms the very basis of that branch of engineering.

[tsk1979]

Originally Posted by Samurai

So... we have maths lecturers in engineering colleges telling electronics and communications engineering students not to bother learning Fourier transforms. I am sure that idiot of a lecturer didn't know that Fourier transforms forms the very basis of that branch of engineering.

Or maybe that brilliant lecturer knew that the candidate would be writing C code and never ever bothering with electronic circuits or signal programming, so did not need to waste time on something which had a big chance of lowering their score?

[NetfreakBombay]

Campus season is here again, from few collages that I have visited so far (For a Software developer job):

1. Average programming capability is way better
2. Students highlight HackerRank / CodeChef / TopCoder scores in ~60% resumes
3. Collages now focus on competitive programming from 1st year onwards (Not sure it is driven by collage or by students themselves)
4. Almost all companies that were visiting, had online coding test as first round (some did away with ANY requirement for Gades/CPA).

Because of this, average students are pretty good at applying algorithms. Since most of these platforms limit both time and memory, students now are pretty good at basics.

None of thee collages are IITs/NITs.

[Samurai]

Originally Posted by tsk1979

Or maybe that brilliant lecturer knew that the candidate would be writing C code and never ever bothering with electronic circuits or signal programming, so did not need to waste time on something which had a big chance of lowering their score?

Such a decision is incredibly stupid. By denying them FT, the maths lecturer eliminates for them any career that requires communication theory. My college did that to me in the 80s. When will this stupidity stop?

I keep asking E&C candidates with great math score about FT, because I need people to work on the echo cancellation and voice analytics. But these colleges ensure I never get anybody with FT. Anyway, I am done depending on these colleges. I am going to ask couple of these freshies with good math scores to learn FT as part of the training.

Originally Posted by NetfreakBombay

1. Average programming capability is way better
2. Students highlight HackerRank / CodeChef / TopCoder scores in ~60% resumes
3. Collages now focus on competitive programming from 1st year onwards (Not sure it is driven by collage or by students themselves)

Is this a regional phenomena or have you seen it across India? I haven't seen it in the local student population. I am still seeing distinction graduates in CSE not knowing how to replace a character in a string.

[NetfreakBombay]

Originally Posted by Samurai

Is this a regional phenomena or have you seen it across India?

Have seen focus on competitive programming in quite a few Mumbai / Pune and Ahmadabad collages.

Surprisingly, Bangalore collages did not do as well in this specific aspect. This is based on just 6 collages in Bangalore so data is not representative.

Originally Posted by Samurai

I am still seeing distinction graduates in CSE not knowing how to replace a character in a string.

I am sure there are CS students that can't code. With coding tests, it very cost effective to screen for hundreds and get 10 - 20 that can code and you want to interview.

Most cumbersome part is arriving at "Difficulty level" of tests. In my experience, it takes 2 - 3 years of internship + placements seasons to arrive at appropriate level of difficulty of tests.

Positive side is, once done; it works across the globe with minor upkeep every season.

[Samurai]

Originally Posted by NetfreakBombay

Most cumbersome part is arriving at "Difficulty level" of tests. In my experience, it takes 2 - 3 years of internship + placements seasons to arrive at appropriate level of difficulty of tests.

We give 5 rounds of coding tests of increasing difficulty.

Originally Posted by NetfreakBombay

I am sure there are CS students that can't code. With coding tests, it very cost effective to screen for hundreds and get 10 - 20 that can code and you want to interview.

Since 2010 I have given up on coding tests as an eliminator. In fact, only one selected candidate since then has passed the 5 level coding test. I realised that the reason for failing the coding tests lies with the college and their incompetent lecturers. If the lecturers can't code, what can they teach to their pupils.

We still have coding tests, but one can advance further despite failing the coding tests. If the person does well in communication test and interview, where he/she displays technical ability and an open minded curiosity, we discount the coding test result. Don't ask me how I judge that, I have developed an ability to make subjective judgment of these factors among fresh graduates. That gamble has paid off, I have turned most of those initial non-programmers into pretty good programmers.

[NetfreakBombay]

Originally Posted by Samurai

We give 5 rounds of coding tests of increasing difficulty.

That is interesting. Are these part of the same test.. or do you eliminate in earlier rounds ?

Originally Posted by Samurai

Since 2010 I have given up on coding tests as an eliminator. In fact, only one selected candidate since then has passed the 5 level coding test. I realised that the reason for failing the coding tests lies with the college and their incompetent lecturers. If the lecturers can't code, what can they teach to their pupils.

Yes, in most collages pay is not sufficient to hire a teacher who can code (exception do exist, but they is just that.. exceptions).

What I have noticed is:

1. Current batches are "digital natives", have more access to computers and at earlier age. In 1991, only 4 - 5 students had programmed before B.Tech in my batch.
2. Peer pressure to get into good companies and the knowledge that coding is a must.

Clearing this round means a shot at getting a job of 8 - 15 lakh in startups / product v/s 2 - 4 lakh in IT Services. This I believe is a big motivator.

http://placement.daiict.ac.in/home/placementStatistics

In this collage that I visited, all companies paying 7 lakh or more expected that student would be a pretty good programmer and problem solver and would have worked in startup/corporate environments in internship.

From a batch of ~300 students, ~60 got such jobs.

[Samurai]

Originally Posted by NetfreakBombay

That is interesting. Are these part of the same test.. or do you eliminate in earlier rounds ?

If the candidate solves the first program with difficulty level 1, we give another program with difficulty level 2, and so on.

I am glad hear that in some colleges they have started trying their hand in competitive programming. A decade ago only top colleges like IIT were in on it.

We tried using topcoder to test the candidates a decade ago, not a single one got through.

Damn, I have not logged into my topcoder account in 10 years now, not after participating in Google Code Jam 2005.

[Octane_Power]

Originally Posted by NetfreakBombay

Yes, in most collages pay is not sufficient to hire a teacher who can code (exception do exist, but they is just that.. exceptions).

Originally Posted by Samurai

If the candidate solves the first program with difficulty level 1, we give another program with difficulty level 2, and so on.

Good ideas being shared there! However I have this one question and would like to know your views on it.

Top companies (especially in US) ask very difficult algorithm solving problems in their technical/coding interviews. Why is this? 7/10 times the algorithms are quite tough and (I feel) also irrelevant.

I remember reading on Quora where a senior exec/developer of a tech company wrote that even he doesn't get this idea of asking 'tough algorithmic questions' while hiring. Because, in real life software development, algorithms aren't the main concern. Good design is the main concern.

OK! Now I understand, asking tough algo questions is not just for testing the candidate's algorithmic knowledge but in the process its seeing how he/she goes about the problem, thinks of the solution, breaks down the problem etc. But I feel they should be upto a certain fixed level of complexity, not super difficult. Some more importance should be given to product designing, domain knowledge too (this is for candidates with some experience, not freshers).

-Bhargav

[NetfreakBombay]

Originally Posted by Octane_Power

Top companies (especially in US) ask very difficult algorithm solving problems in their technical/coding interviews. Why is this? 7/10 times the algorithms are quite tough and (I feel) also irrelevant
..
..
..
But I feel they should be upto a certain fixed level of complexity, not super difficult. Some more importance should be given to product designing, domain knowledge too (this is for candidates with some experience, not freshers).

There are few competing priorities that interact here (based on my experience, as well as fair dose of extrapolation).

1. In India, campus recruitment is done in a compressed timeframe. In some collages, Yes/No must be decided in just few hours. This leads to "Err on the side of caution". Cost of a bad hire outweighs risk of not selecting someone good.

In US schools, one would get to spend time on each interview and look at all aspects and not just specific kind of problem solving.

2. At the time of Hiring, it is not known if someone would join teams that work on core systems or teams that work on API based integration. So Hiring is done based on the fact that he/she might have to work on either.

3. "running the gauntlet", both parties know that this just a barrier for entry. If someone can work hard to get through this barrier, she has both capability and willingness to be successful in other things.

4. Some companies can afford it. Their "pull" is so high that they will get people anyway.

[Lalvaz]

Originally Posted by Samurai

This time I came across another candidate with insane score in mathematics in every semester. Since she was an electronics and communications engineering graduate, I had to ask...

Me: Do you know Fourier Transforms?
She: No
Me: But, your scores in mathematics is too good, how did you not learn Fourier Transforms.
She: Our maths lecturer asked us to skip it in choice. Even without studying FT, we could aim for 100 percent marks.
Me:

So... we have maths lecturers in engineering colleges telling electronics and communications engineering students not to bother learning Fourier transforms. I am sure that idiot of a lecturer didn't know that Fourier transforms forms the very basis of that branch of engineering.

While I'm sure Fourier transforms must be essential in your line of work and it would obviously help if the recruit knew it beforehand, is it so difficult for someone to pick it up?

Is is worth rejecting an otherwise intelligent person for not knowing FT?

[Samurai]

Originally Posted by Lalvaz

While I'm sure Fourier transforms must be essential in your line of work and it would obviously help if the recruit knew it beforehand, is it so difficult for someone to pick it up?

Is is worth rejecting an otherwise intelligent person for not knowing FT?

Where did I say rejected them on the basis of not knowing FT? I have selected many E&C candidates in spite of their ignorance of FT, based on other positives. In fact, I just mentioned couple of posts back that I am going to train them on FT since they don't know it already.

PS: I actually hired the person in question. But I mention it here because people (including college staff) don't realise the importance of maths to engineering. Without FT, you are not really a communication engineer.

[tht'snotME]

@samurai

the problem is with the IT companies and recruitment policies.

please dont take me wrong. How many you recruit per year compared to the MNC's which recruit atleast in multiple of hundreds from a same college. their only consideration is % .

The pressure one faces from their family/relatives for not getting the cutoff percentage is huge in current scenario.


In current generation it will be like this i believe. i hope it changes and whole recruitment should be on once capability than their percentage.

i have an interesting example to share
i know a person, who has dropped out of engineering because he cant take the pressure of exams. But he is super genius in C. He works for a startup with the IITians, do you really think th MNC's will notice him? i think the chances are 5 % and less.

[Lalvaz]

Originally Posted by Samurai

Where did I say rejected them on the basis of not knowing FT? I have selected many E&C candidates in spite of their ignorance of FT, based on other positives. In fact, I just mentioned couple of posts back that I am going to train them on FT since they don't know it already.

PS: I actually hired the person in question. But I mention it here because people (including college staff) don't realise the importance of maths to engineering. Without FT, you are not really a communication engineer.

My bad, I guess I missed that post of yours. Its great that you're now training these candidates in the core skills that you require. However, you might also want to consider speaking to the principal of the colleges that you usually recruit from. From my experience, college principals are generally very open and receptive to suggestions/feedback from industry professionals, and would usually welcome the idea of better educating their students. This would help you in three ways, namely, (1) Increase the pool of resources knowing FT, (2) reduce the learning curve for the students, (3) reduce training costs for you.