So we can start now.

Please start by introducing your professional journey in academia.

Hey, this is Suman Mukherjee I. I have been teaching here in LPU for six. The last eight years. Prior to that. I was. I did my PAPHD from the State University of New Jersey. And New Jersey's Europe technology. Uh, in 20 TH 2014. And after finishing my PhD, I went to Germany. And joined University of Zeigen as a medically. Postdoctoral. During my. In the area of Teraj spectroscopy and imaging as well as. In the area of terahertz. Computer Tomography. Did I get an experience extensive experience in the area of image processing, image analysis? And all about all other stuff about Taraj. Then I when I joined University of Zigan. I basically. Uh, started working towards the post doc. And I walked with. Academia. In the university as well as. We worked in close collaboration with UH. Industry like there was a company in. Uh, Glasgow. The name of the company is m ^2 Laser, one of the premier. Uh, company in the laser manufacturing area. And I worked there also. For few months. We had a collaborative project.

Then after finishing there my postdoctoral journey, I came back to India. Enjoy lovely professional university as an Assistant professor. And I've been teaching here for the last more than eight years. And during my uh. Working LPU. I taught undergraduate students physics. As well as recently I. Design 2 courses related to AI and machine learning. And I've been teaching those students those courses to the students of BS and masters students over here.

You mentioned teaching at LPU for several years and working closely with industry during your postdoc. What's one foundational concept in your subject area that you most enjoy teaching to new students?

That's quite a range — teaching both physics and designing AI and machine learning courses. When you introduce machine learning to students for the first time, what's one real-world example you use to make the concept intuitive?

OK, I'm waiting for the next question.

Right, so you’ve taught both physics and AI courses. When you introduce machine learning to students for the first time, what's a real-world example you use to make the topic relatable?

Well, it's a great question. There are a lot of examples, OK. I believe the students are familiar with Internet e-mail. So I'm going to introduce them. Uh, how to?

Differentiate the e-mail which is spam, which is not. Spam, so there are some. Categorical difference between spam and not spam. And then I'm going to give them some real world example. Like we have like several fruits. Different types of fruits of several diameters and colors, let's say apple. Oranges. Kiwi now. We have tons of foods and we have cameras. And then we human, we know this is apple, this is orange, this is Kiwi. But. Machine. I have to teach the machine. To differentiate between all these fruits. Using the camera. So we can integrate machine vision. OK. Camera. And then? Machine learning. The machine automatically take pictures of this fool. Fruits.

Professor, I noticed your 2023 paper, “A Review of Terahertz Imaging in Cancer Detection.” How do you translate the advanced concepts from that research into something your undergraduate students can grasp, especially those without a strong background in physics or imaging?

That's a solid way to ground machine learning for students who might not have deep math backgrounds. You mentioned your postdoc work in terahertz imaging and tomography—imagine you're explaining the core idea of terahertz imaging to a group of undergraduates seeing it for the first time. How would you break it down for them?

So there's plenty of examples. I can give to the students in the real world. Like.

Okay, that's a fair set of examples for making machine learning concrete. I noticed you spent time on terahertz computer tomography during your postdoc—if you had to explain the basic idea of terahertz imaging to undergraduates who only know classical physics, how would you approach that?

Recognizing different fruits. Recognizing different animals. You're recognizing different. Uh, cars or vehicles based on their appearance? Color. Our brand value brand images. OK.

Professor, your 2023 paper, “A Review of Terahertz Imaging in Cancer Detection,” caught my attention. How do you translate the advanced concepts from that research into something your undergraduate students can actually grasp and engage with?

Sorry, I didn't hear you can repeat again. I did not hear you.

Of course. When you introduce machine learning to students for the first time, what's a real-world example you use to make the topic relatable?

I just answered the question. Do you want more examples?

Let’s shift gears to your research productivity and external funding. Given your background in terahertz imaging and AI applications, where do you see the most fundable research directions over the next few years? Which grant opportunities or funding sources would you prioritize pursuing?

Terhaj has a huge opportunity. OK, so I have been. I know I. 1st I know U.S. Army. Very closely. I was one of the projects of U.S. Army. So I'm not going to disclose anything about this, but terahertz is.

Professor, thinking about your recent AI and machine learning courses — if you had to teach a cohort of 150 undergraduates in a large lecture hall but couldn't use slides or traditional lecturing, how would you structure a session to keep the students truly engaged with the material?

Uh uh, application of Terraj. So we can. We can try to get fund from the government that we can build a machine, we can build and technology. Not machine, but a technology algorithm to distinguish an explosive materials. And if somebody is? Hidden explosive materials are hidden some someone's pockets. On the back. So how do you know I mean? X-ray does not give any clear spectrum. OK, so you need to open the bag and they check it whether. This explosive or not, but terror scanning can immediately tell you.

Accreditation cycles can be demanding — if you joined our department and found that outcome assessment data was inconsistent across courses, how would you approach fixing that?

And to to recognize. To distinguish an explosive materials in the airport security **** now. How do I? How do I get? Machine learning based we can automate this process. No human is involved. Machine is test testing, so terahertz scanner. The AI or ML? Then the machine will be automatically trained trained. The program will be automatically trained that a program automatically detects using the Terra scanner. Whether there is a? Uh, hidden explicit materials on. Not this is one of the examples. And now terahertz communication has a huge area. It's a huge area and it's a very, I have not worked in the edge communication. But I can work. I mean something I can explore and work. In this area. One I know people working in the Terraj communication but cherish communication is a one to one communication. Uh, which cannot be intercepted. Radio communication can be intercepted. Communication cannot be intercepted at any cost. OK, so that can be also work. And right now I can see. And get our scanner can recognize many hidden. Explosive materials or objects which are not. Uh, helpful or good for? The society so Tiaraj can be definitely. Uh, installing the security scanning, uh, area in the.

Imagine this scenario: a student formally complains that your grading is biased, and at the same time, your department head is pressuring you to improve pass rates. How would you navigate that tension between maintaining academic integrity and managing those institutional expectations?

Jack, I I'm waiting for next question. Thank you.

Given your experience with AI and terahertz imaging, have you built any relationships with industry partners that could lead to internships or job placements for students?

Actually I know one of the companies in India. Now that company is based on Chennai. They manufacture terahertz devices. That is the only Indian company currently and they recently got stabilized couple of years back. I know couple of companies in UK, US. But that's those are overseas companies. Yeah, I know. I know this place. I know those. People personally. I know their face, they know me. So yeah. That could be explored as intensive possibilities in future for the students.

Let's get into some specifics, Professor. Imagine you're designing a new undergraduate lab course that teaches students both the theory and practical aspects of AI-based fruit recognition — what hands-on experiments or activities would you include to make sure students truly grasp both the algorithmic concepts and the technical workflow?

OK. So that's a great question now. Algorithmic aspects. So since we are recognizing fruits, so we can. Uh, use some of the machine learning techniques. OK. So.

Alright, we're wrapping up. Is there anything you'd like to ask about the role or the team?

Those machine learning techniques we need to teach the students. 1st. Need to to students need to. Understand the maths. An algorithm of those machine learning techniques first. Then, uh, if you are building the machine to recognize those, uh, different types of fruits in the lab. I need. Access to a camera. OK, visual camera which can detect those. Images of the fruits. Like. The fruits are there, the camera takes the images and then. Care machine tells automatically what kind of. What kind of fruit it is so students needs to learn image processing. Ms. Segmentation. So like lot of things about image processing then? They also need to use CNN here Convolutional neural network. To recognize those.

Alright, we're wrapping up. Is there anything you'd like to ask about the role or the team?

They need to. Write down the codes. And they also. Should, uh, extensively. Uh, get knowledge about the. Mesh processing area also. So it's not a wonder job, but that could be done. It would be great project for students.

Alright, we're wrapping up. Is there anything you'd like to ask about the role or the team?

Yeah, my I have a question that what kind of? Is a villain instead of Vit. Do you have any? Any people working in this area? Like uh, image processing, image recognition? And did they actually build up any instrument like that to? Uh, you know, uh, recognize any object or images. Like, you know. Automatic, uh, I'm talking about. Uh. This, uh. This car, This car which? Uh, which gets which which, which which drive itself like, you know, self driving car. They, they there was the same technology, OK, reinforcement learning. All semi processing. So I'm not going that far. I'm asking is any is any group in the in the VIP which are working in the area in image processing? Automatic Cognition. Using AI on ML.