CDS iisc

How do I prepare for IISC’s CDS (Computational Data Science) interview for M.Tech (research)?

EDIT 2: CDS department is conducting pre-interviews which can be helpful to look into for candidates short listed.

EDIT 1: I did my masters in Computational Science stream, the bias should be evident in the way I explain the interview process below. So take the advice with a grain of salt.

Here is a summary of personal experience gained from interviews over three departments in IISc

General Tips:

  1. Interview typically goes about as a test on knowing what a possible solution could be to a problem (typical questions on basics like definitions, concepts, etc). The benefits of these questions are two folds
    1. Makes the candidate comfortable.
    2. Gives a quick assessment about the candidate to the committee on the clarity in fundamentals, based on the ease with which one can answer and defend some cross questioning to cover corner cases. These starting one or two questions per topic should be easy pickings.
  2. Knowing how to arrive at the solution most of the interview will be dedicated to this stage, also from this stage onwards the questions will be focused on how the candidate responded to the previous questions. The reason being
    1. When s/he has answered the basic questions and constructed examples to demonstrate the corner cases or importance of necessary condition in a definition. S/He is providing insight into how s/he is thinking, the committee is now focused on how s/he can stretch the extent of this understanding.

CDS Interview:

CDS traditionally has two focus areas a) computational methods to handle large data (computational science) b) systems that can implement these solutions (computational systems science). I was told the interviews for these two streams have been merged now. Back when I was interviewed we had to indicate the preferred stream and choice of lab I would like to join within that stream. This is relaxed these days as most of the candidates have necessary fundamentals for both the streams and cannot make an informed choice without being at the department. Choice of the stream is not super important as the interview for both is similar with slightly different focus. Topics important for CDS

Must know:

  1. Linear Algebra, with focus on matrix computations
  2. Programming in C with focus on complexity of the code, and learn how to code in order to optimize in terms of time and memory.

Computation science Stream: If you have made up your mind to pick this stream

  1. Basic Numerical methods learnt in Bachelors : e.g. root finding, operational research (simplex methods, pivots), Gaussian elimination etc
  2. Signal processing: e.g. FFT, DFT, Nyquist sampling criteria etc
  3. Real Analysis: If you have a mathematical back ground
  4. Probability and Random processes

System Science Stream: If this is your focus of interest

  1. Computer architecture: e.g. cache management, processor architecture, memory design and interface etc
  2. Data structures and algorithms
  3. Database management
  4. Parallel processing
  5. Distributed systems

Tips: Usually you are asked to pick at least two topics you have prepared for the interview. Remember that these topics should be used to convince the panel you are a worthy candidate. The panel has two goals in mind

  1. To assess your analytical thinking.
  2. To figure out labs that could be a close fit to your interest and abilities.

For example

Strategy 1 (Strong general competence): You will be tested anyways on linear algebra and programming basics in C. So why not prepare for Linear algebra and Data structures and algorithms you will have a very focused interview on them. The panel can then ask you to pick another application area and if you say Signal processing and do well in this. The panel is most likely to see you as a candidate for Medical Imaging Group, Computational and Statistical Physics Lab (CSPL), Video Analytics Lab

Strategy 2 (Strong application competence): You can have a very strong focus on application area like say Signal Processing and Probability and Random processes do well there and then when panel will test you on Linear algebra you do well there as well . The panel is most likely to see you as a candidate for Medical Imaging Group, Video Analytics Lab, Machine And Language Learning (MALL) Lab

Related Posts: