Any tips on a Skype meeting? (Urgent!)

[Nirvik]

Guys!!

I just received a notification with a request for a Skype interview. I have been asked to prepare a presentation on two topics, none of which I know about right now. The interview is scheduled for this Friday at 4 p.m. PST. Can anyone let me know as what I should be prepared for and any pointers on what I should present?

[morpheus]

It sounds like they want you to present on two projects that you've worked on in the past. What's your time limit? 

[Nirvik]

7 minutes ago, morpheus said:

It sounds like they want you to present on two projects that you've worked on in the past. What's your time limit? 

They haven't given me any time limit as of yet. They did say that one topic will be about something that I know of beforehand. This I am predicting to be Machine learning, as my POI did say that she is looking for someone to teach her students about Machine Learning. The second topic being something that I don't of yet. 

[Nirvik]

23 minutes ago, morpheus said:

It sounds like they want you to present on two projects that you've worked on in the past. What's your time limit? 

So, I was just contacted by my POI and I right about the topics. She said that it's going to be 45 minutes of presentation, followed by 15-30 minutes of discussion.

Since she asked for my transcripts, should I expect any questions regarding my undergraduate performance?

[morpheus]

Oh, so it's an assessment of your teaching skills? There are a ton of good practices to bear in mind. Spend extra time planning the flow of the presentation; begin the presentation with a goal or outline (and refer back to it); and keep at MOST one equation or algorithm on a slide- pseudocode only, unless you need specific implementation details. Time yourself when practicing to get a feel for the correct pacing.

Here's a screenshot of my standard slide layout when I present Machine Learning topics- in this case to a weekly reading group (everyone reads the same 2-3 papers and we discuss together). And for the record- Spatial Transformers don't work in practice  

[morpheus]

16 minutes ago, Nirvik said:

So, I was just contacted by my POI and I right about the topics. She said that it's going to be 45 minutes of presentation, followed by 15-30 minutes of discussion.

Since she asked for my transcripts, should I expect any questions regarding my undergraduate performance?

Is your undergrad performance an issue? I'd imagine that most of the time will be filled by your presentation, and other questions they might have concerning your past experiences/projects. 

[Nirvik]

50 minutes ago, morpheus said:

Is your undergrad performance an issue? I'd imagine that most of the time will be filled by your presentation, and other questions they might have concerning your past experiences/projects. 

I don't think the undergraduate performance would be an issue. Yes, it has its ups and downs. But besides that I hope they stick with my project work and the topics that they asked me to present.

 

[morpheus]

Then they were probably just interested in which advanced courses you've taken. If this were a math PhD they would also ask which textbooks you used in those courses

[Nirvik]

56 minutes ago, morpheus said:

Oh, so it's an assessment of your teaching skills? There are a ton of good practices to bear in mind. Spend extra time planning the flow of the presentation; begin the presentation with a goal or outline (and refer back to it); and keep at MOST one equation or algorithm on a slide- pseudocode only, unless you need specific implementation details. Time yourself when practicing to get a feel for the correct pacing.

Here's a screenshot of my standard slide layout when I present Machine Learning topics- in this case to a weekly reading group (everyone reads the same 2-3 papers and we discuss together). And for the record- Spatial Transformers don't work in practice  

Thank you. This seems like a brilliant presentation. Can you provide some more information regarding what else I could speak about? My first topic is about Artificial Neural Networks. The second is on Deformation registration in biological problems. The second topic contains a lot of equations. Should I try to like summarize them into a few and present?

[Nirvik]

2 minutes ago, morpheus said:

Then they were probably just interested in which advanced courses you've taken. If this were a math PhD they would also ask which textbooks you used in those courses

I see. Thanks. 

[morpheus]

2 minutes ago, Nirvik said:

Thank you. This seems like a brilliant presentation. Can you provide some more information regarding what else I could speak about? My first topic is about Artificial Neural Networks. The second is on Deformation registration in biological problems. The second topic contains a lot of equations. Should I try to like summarize them into a few and present?

With 45 minutes, I would plan to devote 20 minutes to each topic. So I'll consider those two topics separately.

ANN's: Wow- only 20 minutes to give a nice introduction to Artifical Neural Networks! I would give the talk as though the audience had never heard of ANN's before: describe the basic structure, explain why they're useful, and close up with a simple example (maybe MNIST) and famous applications. Luckily for this topic, you have a TON of online resources which demonstrate various ways of introducing the subject. I would look through how other instructors begin their ML courses and choose the best examples from those. Use this part to show off your teaching abilities for intro-level classes (and speak slowly!). 

Deformation registration: Old-school methods or new ones? Here's where you can show off your ability to teach slightly more advanced topics. I stand by the one-equation-per-slide rule: even as a math student I was told to adhere to it. We were told that people in the audience can really only process one equation at a time (with the occasional exception), and that if an equation is important enough to make it in the presentation, you must explain it carefully. So yeah: choose the most important few equations and summarize the general theory first. Then you can show some fancy examples and, if possible, finish off with a single slide on recent advancements in deformation registration, to demonstrate that you follow current research. 

[Nirvik]

27 minutes ago, morpheus said:

With 45 minutes, I would plan to devote 20 minutes to each topic. So I'll consider those two topics separately.

ANN's: Wow- only 20 minutes to give a nice introduction to Artifical Neural Networks! I would give the talk as though the audience had never heard of ANN's before: describe the basic structure, explain why they're useful, and close up with a simple example (maybe MNIST) and famous applications. Luckily for this topic, you have a TON of online resources which demonstrate various ways of introducing the subject. I would look through how other instructors begin their ML courses and choose the best examples from those. Use this part to show off your teaching abilities for intro-level classes (and speak slowly!). 

Deformation registration: Old-school methods or new ones? Here's where you can show off your ability to teach slightly more advanced topics. I stand by the one-equation-per-slide rule: even as a math student I was told to adhere to it. We were told that people in the audience can really only process one equation at a time (with the occasional exception), and that if an equation is important enough to make it in the presentation, you must explain it carefully. So yeah: choose the most important few equations and summarize the general theory first. Then you can show some fancy examples and, if possible, finish off with a single slide on recent advancements in deformation registration, to demonstrate that you follow current research. 

Since I don't have much knowledge about Deformation registration from images in Biological problems, do you think I should start with creating the presentation with this one, rather than the ANN? Since I have 2 papers of the Deformation registration to study and create a presentation with another presentation on ANN. They both need to presented after 3 days.

[morpheus]

2 minutes ago, Nirvik said:

Since I don't have much knowledge about Deformation registration from images in Biological problems, do you think I should start with creating the presentation with this one, rather than the ANN? Since I have 2 papers of the Deformation registration to study and create a presentation with another presentation on ANN. They both need to presented after 3 days.

If it were me, I would devote today to making the ANN presentation as good as possible. That way I could open the presentation with a topic that I feel confident presenting, and I can spend the rest of the week working on Deformation registration without worrying about going back to the little things. But, it's up to you! Whatever you feel is best  

[Nirvik]

5 hours ago, morpheus said:

If it were me, I would devote today to making the ANN presentation as good as possible. That way I could open the presentation with a topic that I feel confident presenting, and I can spend the rest of the week working on Deformation registration without worrying about going back to the little things. But, it's up to you! Whatever you feel is best  

Thank you for the information.