TSQL Tuesday #93 – My Interviewing Experiences

Kendra Little (b|l|t) is hosting this months TSQL2SDAY which is on the topic of:

TSQL Tuesday #93: Interviewing Patterns & Anti-Patterns

What advice do you have for people preparing for or going through an interview?

I got of to a late start, and I apologize in advance for what may be a bit messy blog post.

Background

A couple of years ago, when I was working at a pretty large blueish container shipping company, we got the opportunity to interview candidates for our off-shore team. This was, as far as I know, a first off in the company, and only came to be because my manager at the time was adamant about it.

We quickly decided, that I was to take on the technical part of the interview and my manager the more person/team oriented aspects. So I devised a good number of questions, so I could accommodate for any specialization the candidate may have. We had to cover a lot because our landscape was, well, let’s just put it: Not so single vendor-ish.
To make a long story short, the setup in the company at that point in time was that the Data Warehouse (DWH) was situated on a Teradata box. On top of that there were several options for business reporting. One such option was Microsoft Reporting Services (SSRS), which was done in the department where I was sitting. The reports were however not sourcing directly from Teradata, but from one of many Analysis Services (SSAS) cubes we hosted as well. All data was sourced using Informatica (into the DWH), in yet another department, so Integration Services (SSIS) was officially (and unofficially) not allowed as sourcing tool, which is why no questions reflect that aspect of the Microsoft BI Stack.

So, we were looking for strong profiles in SSRS and SSAS as well as having Teradata knowledge. Along with a general understanding of Data Warehouse fundamentals of course. In other words, not junior profiles.

Preparation

In order to lock in on the level of the one being interviewed I prepared a set of questions with increasing technical difficulty. Since my knowledge on Teradata was limited I asked around among the DBA’s and decided after some contemplating to completely avoid checking for technical Teradata skills. We would have to train them, if need be.
The list of questions evolved during the whole process. I honed in on the  general level of applicants, after firing questions East and West. Me being more precise gave everyone a much better experience. Conversations tend to stop, when no-one understands you and, you don’t look like a Donkey asking about simple stuff.
So hopefully you get the opportunity to tune into the level of the applicant before you unload your tech quiz.

The Interviews

I wound up having three categories of questions:

Proficiency

Do they have any strong sides? Usually people tend to have a favorite tech in which they excel. So for us it was important not to hire only SSRS Wizards/Ninjas or whatever it’s called nowadays. We wanted a fair mix.

I would ask the interviewee to rate them selves in the different tools and languages and use their rating as pointer to the question I would ask on the technical stuff. If they rated themselves > 7/10 in MDX and are having trouble naming at least five (5) frequently used functions, you know something is rotten.

Something I notoriously have gotten wrong would be:

  • Q: Explain what you get from having Enable Visual Totals checked?
  • A: When Enable Visual Totals are checked, the Grand Total reflects the total you are allowed to see. The Grand Total is not a Grand Total, but your Local Total. At least that works in my mind!
Data Modeling

Data modeling was usually the area where the interviewee’s were least knowledgeable. So the nature of the questions in this category was on a very high level.

  • Q: What’s the difference between a data mart and a data warehouse?
  • A: A data mart is a subset of the data warehouse.
  • Q: Do you know any architectural principles of data warehousing
  • A: Inmon & Kimble
  • Q: Delivery Date and Shipping Date in a Fact table, how would you model that?
  • A: I would use a role playing dimension, Time.
Optimization

In this category of questions, I’d ask them to explain how they would optimize say a poorly performing dimension in SSAS.
What about query performance tuning – what would they look for and how would they attack the issue?

Final Question

But, the one question I always looked forward to was:

Tell me about a Business issue you solved, which you are particularly proud of.

This question I grew fond of, because it allows for the interviewee to brag about how cool they were doing this and that awesome thing to save the business a Centillion Dollars. This allows the person really to shine and at the same time, you get a feeling for their level of engagement and technical flair, not their skill set as such. Don’t you want to know how ingenious they can be a their best? This will give you a hint in that direction, or at least it did for me.

Tests

Back in the days before the big blue container hauling company, I was also involved in the hiring process. I was in fact tested myself, when I first began working there and in my experience tests can have two immediate take-away’s:

  1. It was all talk, and no walk.
  2. Does the person think and act like you expect them to?

1)

I’ve had a number of people skipping the test after we had a good 1-1½ hour talk. Excuses flying around like butterflies in the summer.

“Oh, I forgot! It’s my Grandmother’s birthday, I have to leave urgently. Bye Bye!”

“My car is double parked… I need to move it immediately!”

“Is that the time?”

– Never to be heard from again.

For this kind of situation, a test is very handy – apparently there are a fair amount of bul$hitters out there and fortunately the test prevented you from hiring them.

2)

In my mind, a test should display how the person got to the result, rather than just showing a number. I want to see the T-SQL code, the MDX, M or whatever they used to solve the puzzle. This tells me way more than a result and time.

With the actual code in hand, you can go into discussion (something a lot of coders dislike) about how they came up with what they submitted as final answer. By discussing the code, you get a sense of how their personality is when dealing with their own code. Expect a lot of back-paddling, as time, room temperature, humidity and other factors can have a high impact on the delicate nature of some coders 😉

Summary

There are a lot of ways to conduct an Interview, and I bet my humble contribution to this tsql2sday is one of the more novice ones, but I hope I have allowed you to see some of the thought processes I’ve had, when having to interview people for a position. I am very much looking forward to seeing the other contributions to this blog party – can hardly wait.

Thanks Kendra for hosting!

One thought on “TSQL Tuesday #93 – My Interviewing Experiences

  1. I really enjoyed reading how you designed the interview process for that broad range of topics — that’s no easy challenge. I think my favorite part is that you pointed out how you purposefully made the final question something that could end the interview on a high note — and with an interesting story. That is a great suggestion, and something I never thought of. No matter how an interview goes, it is good for everyone to try to structure the interview to end with a positive story from the canddiate. I’ll remember that!

Leave a Reply