Yes, we can get into.

Please start by introducing your professional journey in academia.

So, yeah, thanks for this opportunity so myself. So I'm currently working as an assistant professor at Value Maintenance College in Chennai. So I have around, uh, two years of academic experience. An industry experience, uh, I mean, plus, uh. Umm, industry experience. So. Umm, currently I'm handling some just like operating system and virtualization and I'm also. And guiding students on LLM and AIM projects. So. Yeah. That's about.

You mentioned guiding students on LLM and AI projects — what do you find most rewarding about teaching those subjects?

We get to. Learn all the shortcomings along the way along with the students and UH.

More recent trends on AI models.

And enlightening them.

You presented your paper on "Biomedical Image Denoising Using CBDNet and U-Net" at the 7th International Conference on Engineering Research and Innovations. How have you drawn on that research experience to make the topic accessible when teaching undergraduates — especially those who aren't familiar with deep learning or medical imaging?

That's pretty high-level — can you give me a concrete example of how you helped a student understand a tricky concept in AI or LLMs?

Uh, real-values cases.

Right, so that's still a bit general. Let's shift gears. I noticed you presented a paper on "Biomedical Image Denoising Using CBDNet and U-Net" at the 7th International Conference on Engineering Research and Innovations. How have you drawn on that research experience when teaching foundational topics to undergraduates?

So this was actually part of my M Tech thesis work. Uh, my final year project so. So at that time I was working on. Medical image denoising so using CBD net and unit. So. So the work was mostly towards medical images denoising so more. At that time, most of the at the time in the year of 2022 or 2021. Image innovating works for mostly inclined towards uh. Uh, general use case images and not medical images so. My goal was to see if the general use case algorithms worked for the medical images. And, uh, to see how well it performs. So at that time. I uh, we do. Umm 2. Medical image data set 1 was the. Text. And the brain MRI chest X-ray was a real world data set obtained from open data sources. And the brain, Brain. Brain data set. Avrene MRI data set was obtained uh through stimulated websites. And then we had. Uh, given varying no noise noise intensities levels of 3/5 and. 9 Percentage of noise intensities of Ricci noise and noise on noise. And then? So first, uh, first motive was check if the. Algorithms of CP linear and new network CBD net was umm. The normally is conventional. Blind denoising subnetwork. Which is a general use case, uh, images. And then unit. Umm, so at that time it was being used for segmentation task. And also there were a few works related to. Uh, denoising images? So. So we we use this to algorithms. And implemented for those medical image noises and based on the results obtained. We use the metrics of PSNR. Um, SSIM. And YouTube University Universal quality Image index. Uqi. And then using this three, uh, three parameters, we checked how well it performed. And how was the computation time?

Interesting, sounds like you designed and evaluated a pretty hands-on research project. When you've taught undergrad students about neural networks or image processing, how have you adapted complex topics from that project to make them accessible in an introductory class?

To be honest, I haven't got an opportunity to. Umm, tease this hands on uh, subjects. Umm, I only got an opportunity to teach NLP and uh. Machine learning, uh, subjects. This was part of my M tech umm. Mtech Finally a project.

That's clear, thanks for being straightforward. In the NLP and machine learning courses you've taught, what's one specific technique or example you've used to make a challenging topic understandable for students with little background in the subject?

I'm not able to recollect it at the moment.

Given your recent work in biomedical image denoising and your interest in AI, where do you see the most fundable research directions emerging over the next three years? What kinds of grants or external funding sources would you target to support that work?

Which are being done in vit. I will try to incline towards it. Work was contributing their projects and this work.

Let's turn to your approach in the classroom. Suppose you're teaching a large introductory AI class — say, 200 students — and you're not allowed to use slides or traditional lecturing. How would you design a session that gets students actively engaged and really wrestling with the material?

Which they will brainstorm. Brainstorm the activities given and we can try to. Slowly bringing the subject into picture.

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?

So I'll try to approach my heads or peers and I'll. We can't, uh. Demand them to be. Demand them to be incorporated. But we can suggest our ways. And ways in which it can be changed. So which will be a better one for both the students and the faculty?

Imagine this scenario, Professor: A student formally complains that your grading is biased, while at the same time your department head is urging you to raise pass rates due to institutional targets. How would you navigate the tension between maintaining academic integrity and responding to departmental pressure?

So. This can be, uh, done. Uh, this, this can't be forced upon students, but we can, uh, course the, uh, take special care for the students. And make sure they score, uh, score well. Uh, make, make the make sure, make sure that we can, we can make them score well.

Through special coaching. Maybe, maybe, maybe we can focus more on the. Weaker section of students. Uh, maybe we can give more, uh, special coaching. That's what we can do.

Hmm, that's pretty vague — what would you actually do if your department head insisted you bump up pass rates, even if you felt it compromised grading standards?

Let me shift focus. Given your background with medical image processing and AI, do you have any industry connections or collaborations that could help students access internships or real-world project experience?

Umm, to be honest, no. Um, because I'm, I'm into the teaching industry only for around 1 1/2 years, so I'm here to get an industry connection. Or, uh. Academy Collaboration. So I'm in my starting face in my career.

Let's shift to your technical depth, Professor. Imagine you're guiding a group project where students want to use AI for anomaly detection in IoT sensor data from smart homes. If their models are underperforming, what's one advanced method you'd suggest to improve detection accuracy, and how would you help them implement it?

So training with the. Training with more details will be the first thing because. Umm, if you're targeting using one home. The the data data set or the parameters would be. So I'll first recommend them to. Uh, increase the. So we should not be fully focusing on one homes, we should be focusing on. On other homes also.

So inserts the data set can be increased in the model will perform well.

That's a reasonable starting point. Now, when you're supervising student research, how do you help a student who keeps hitting dead ends and is feeling discouraged about their project progress?

That's my first. Uh, recommendation to them? close.

That's a reasonable starting point. Now, when you're supervising student research, how do you help a student who keeps hitting dead ends and is feeling discouraged about their project progress?

Right, let's touch on your teaching approach. When you're running lab courses, how do you make sure students who are struggling with the practical parts still learn the core concepts, not just the procedures?

So. Normally when, uh, when we start with the lab course, I make sure they are clear with the. Theory. Other core concepts. So first I'll just we would have already covered them in classes. If the concert hasn't been covered, I will take around 10:15. Message cover them in the classes. And then I will individually, uh, interact with those students to make sure they're able to understand the concepts if. If they aren't. I will teach them for a small small batch or maybe indu individual attention. Then if they're. Well aware with the concepts. They can focus on the practical aspect of how to frame how to. Uh, do the particular project, uh, experiment or the exercise.

Okay, so if a student keeps hitting dead ends and is getting discouraged about their research project, how do you help them get unstuck without simply giving away the answer?

Just give. Giving a simple motivation. And and giving them with an example. Might help.

Alright. Let's talk about your research publications. Can you give a concrete example of one of your published papers, and walk me through your main contribution to that work?

So to be honest, I haven't done any publications. So I only done 1 conference proceedings. And three publications. Which is in it's with. With collaborator with the students.

Alright, Professor, let's turn to student evaluation. When you're setting exam questions or grading, how do you make sure your process is fair and clearly aligned with the learning objectives?

Around 3 papers. So this is, uh, with. Part of collaborative student. I haven't don't have any. Publications as such.

Got it, so let's pivot to your experience with student evaluation and exam duties. When you're setting exam questions or grading, how do you make sure your process is fair and clearly aligned with the learning objectives?

Would you mind repeating the question once again?

Sure. When you're creating exam questions or grading, how do you make sure your approach is fair and that everything lines up with the learning objectives for the course?

Umm umm. Through internal uh. Maybe internal tests or assignments?

Which will make us understand whether they are clear. And while evaluating. We make sure the key answers and key points are mentioned. And not there based on stories. So even if. The answers. It's not straightforward. We will see if they're able to get the crux of the concepts.

Alright, sounds like you use internal tests and assignments to check coverage. What about when you have to grade a project with a lot of subjective elements—how do you make sure your marking is consistent across different students?

Alright. Shifting to your communication and teaching style—when you're introducing a complex technical concept in class for the first time, what's your go-to strategy to make sure students of varying backgrounds can follow along?

And if we as a person or me as a faculty, I'm not able to do the. Uh, dots. I will make sure to prepare in such a way. Uh, so. My IC I'll be able to communicate those complex complex in a simple manner. By connecting the dots with the day day-to-day activities. Or the activities is both. The students and UH students and the. Faculty as myself get connected with.

Alright, let's touch on your doctoral work. What was the core research question of your PhD, and how did you go about addressing it?

I haven't done my PhD yet.

No problem. Let's switch to industry experience—have you been involved in any projects or consultancy work where you applied emerging technologies like AI, IoT, or cyber security in a real-world setting? If so, can you walk me through a specific example?

So do you remember the UH Biomedical MSD nursing project? Which we spoke upon.

Right, you mentioned the Biomedical MSD nursing project earlier. Can you explain specifically which emerging technologies you applied in that project and what your role was?

Access with the tools. But with what? What kind of tools which we have? They gave up. Give up some advice and how to proceed with it. We were, we were mentored. For the project.

Sounds like you got some industry mentorship but limited hands-on access. Given that, how did you make sure you still delivered meaningful results with the tools you had?

Yes, so I had a SO. Main tools which are required as a programming language Python. For that I use the tool called Jupyter Notebook and Anaconda. And then? I actually had a dedicated GPU myself, so mine was uh. 3060 uh RTX laptop. So I didn't have to face any computation issue or computational lag. And then, uh. To so for further for the image processing and for the denoising. So we had. We had the Python library to help us with. So the main tool. Which was needed was my uh, which was storage and GPU because my data set was around. Uh, I don't want to file. Close to 500 MB and I had a GPU of 3060. So, which didn't. So which helped me run the model for quite a longer time and get better results. So, which helped to determine the computation time? Umm, and the better performance.

And measure them using different metrics.

So I just had the 2. Yeah, so I just had the tool, uh. Tools but. That, uh, maybe that industry level tools, we didn't have any access to that. And I think this umm. For this project my. My tools were sufficient.

Okay, sounds like you leaned on Python, Jupyter, and a strong GPU to handle the image denoising project, even with limited access. Shifting a bit—when you're guiding student projects or research, how do you help students define a research question that's both challenging and actually doable with the resources they've got?

Okay, sounds like you leaned on Python, Jupyter, and a strong GPU to handle the image denoising project, even with limited access. Shifting a bit—when you're guiding student projects or research, how do you help students define a research problem that's actually feasible for them to tackle within their resources and time frame?

Okay, sounds like you worked within those resource limits and still pushed the project forward. When you're guiding student projects or research, how do you help students move from just following instructions to actually thinking independently and owning their work?

So normally when we are guiding students. Some students have the capability to build, uh, great projects on their own. Based on their. Well, was technical knowledge. When Umm. Whereas compared to some students where they even don't have a basic. Understanding of what the project is. How a project should be broke. They are my their mind mind will proceed to approach and where they can just. Control C and Control V and do it uh, don't using A A models HR GB T. So. What I will do is I'll just ask them. What is their interest? And if they're not able to put forth an interest on their own, I'll recommend some projects, which is not, uh. Which might not be complex, but involves a lot of work. Now we have tools so. Note the tools won't build a project completely. They have to. Do all these, uh, debugging and all those stuffs and just. Use AI as a tool to help them out and not do it as a full project.

Makes sense—so you tailor your guidance based on the student's background and encourage them to use tools as support, not shortcuts. When it comes to evaluating students, beyond checking key points in answers, how do you ensure fairness and consistency across different batches or sections?

So no since since I'm working in an autonomous engineering college. So there. Uh, our evaluation, uh, is not. Uh, we won't be evaluating. The students we are handling with. We'll be handling. Uh, we'll be evaluating the papers of other class, other Staffs, uh, classes. So if I'm correcting a paper. Umm, I'll be only be able to see the answers, uh, keep points and whether the. Particular, uh. Ancestor diagram, so we'll be given an answer key. So the first good thing is we don't know who the student is. And. So second so. Since we don't know, the student will be only be focusing on the answer keys, diagrams and whether it's. Presentable. But it is a presentable format, so. Like we can read and correct. So I normally I don't. Uh, just seeing the and, uh, key points. I have this. Have a full read through the answer. So to make sure the. Valuation is. So the valuation I'm doing is fair. Without reading and just seeing the key points, it won't make sense.

Right, so you focus on anonymous grading and make sure to read through answers for fairness. Shifting gears—how do you structure your lectures or lab sessions to help students with different learning speeds stay engaged?

Let us talk about the lecture sections so. My typical preparation would be like, uh. A day or two days before. Uh, and then I'll just. Plan the topics ahead like. What enroll? The topics can be covered within that particular period. Uh, 45 minutes or 15 minutes? And. Since I'm. More since I'm off. Just mere I'm just of their age, like a mere age difference. I know student mentality. And I can umm. So I will think in umm. As student shoes like. How will they feel when they teach a topic? And since my son, I have also studied the topics before. So I will know. What, uh, how will the student? Face while sitting in an, uh, sitting in the class. And how to make it interesting? So my my thing is, I'll be framing. The uh PPT in such a way. It doesn't make them boring and before start I'll just not directly jump into PPT. I'll just make sure, uh, we'll start with an example, connect the dots and then bringing them into PPT. If it's not a, if it's not PPD, I'll try to bring it like a. Umm, two way conversation of uh, making sure like they, they will be able to save. Uh, what the topic is and how to? Proceed forward.

You focus on making the material relatable and interactive, which helps. When you teach lab courses, how do you balance covering the theoretical concepts with giving students enough hands-on practice?

Creating a concert.

Can you just repeat once more than I can just start?

I think I lost your thread—when you said "creating a concert," did you mean you try to integrate theory and practice by designing a coherent lab experience, or something else?

Sure. When you're teaching lab courses, how do you make sure students get enough hands-on practice while still covering the theory they need?

So normally, umm. Yeah, so. May I make sure their concepts are clear and? The theory concepts are clear. So when they're jumping into the lab. There will already be having the exercise or the lab manual. So I'll ask them to first try them out. And then see what mistakes they are able to make it. The they're able to. What? What mistakes am I doing? And then based on that, I'll ask them to solve it on your own. Because. Just uh, telling I don't understand the code or don't understand the concept. Doesn't make sense. It's. Umm, the main thing is we have to. Make sure they're able to build a logic.

I mean understand the concept and then based on the concept to build a. Logic or. Code in particular exercise. If, then they are not able to understand how what does the particular line signify and how they're not able to do it. They can ask us.

Right, you want them to try, make mistakes, and build their logic before you step in. Shifting to your research—can you walk me through a publication you're proud of and explain what made your contribution original?

Or if they're doing it in home, they can. Practicing and ask you next day or now we have even support tools which is available.

Right, you want them to try, make mistakes, and build their logic before you step in. Shifting to your research—can you walk me through a publication you're proud of and explain what made your contribution significant?

Right, you want students to understand the concept first, then try things on their own, and reach out for help if they're stuck. Shifting to your research—can you walk me through a publication you're proud of and explain what made your contribution unique in that work?

To be honest, I don't have any publication. So only with the help of student publication, yes. Maybe I can elaborate, Uh. On that if needed.

That's fine—let's talk about student publications then. How do you guide students through the process of identifying a research question and turning it into a publication-worthy project?

First, umm. When they, uh, come and approach me for the first time, I'll just ask them what to say an interest. So as. AI is the. I say it's a Gen. AI now. That's the student. They will mention with first normally. Then I'll ask them to explore what work you are. What? Uh, what you want to do, whether you're strong and, uh, quoting part or whether you're doing, interested, doing and, uh. Image processing at. I'll ask. I'll just. Talk with them to understand what. What is their idea? What kind of process they want to do? And then if they say. Particular topic I didn't ask them to. If they are, if they need a. Problem statement. I'll try to head them out, or if they are interested in a broader statement, I'll guide them. I'll just guide them like. Ask them to explore. Ask them to suppose you're searching up on a particular like. Uh, let us say denoising. What I'll ask them to do is I'll ask to. Take recent 10 publications of 15 publications. Reason of last year, so the current year. And ask them to learn literature survey. First you have to tell them what is the literature survey. Like umm. Like, uh, what is the, uh, what? What are they saying in the papers? What is the current project outcome? What is the drawbacks? And then? After doing our research of 10 to. Between papers, they'll be able to know. What is the drawback and whether we will be capable of doing it with the resources we have?

So you have them do a literature survey to spot gaps and assess feasibility—makes sense. Switching gears, Mr., can you tell me about your experience working on industry projects or consultancy? What role did you play and what was one concrete outcome?

To be honest, umm, I only have worked in one which is the biomedical image denoising umm, which is also as part of my uh. Data Science internship in L&T technology and Services. And which is also part of my Mtech. Finally a project. So do you want me to repeat that again? Because I've already said that to you.

No worries, just to clarify—can you briefly describe your role in the biomedical image denoising project at L&T, and what specific result your work led to?

I can't hear you.

Just to restate my last question—can you briefly describe your role in the biomedical image denoising project at L&T, and what specific result your work led to?

What specific?

Just to restate my last question—can you briefly describe your role in the biomedical image denoising project at L&T, and what specific result your work led to?

What specific results have worked on? The project.

Let's shift focus—how do you adapt your teaching or communication style when a student isn't grasping a concept, even after your initial explanation?

India faces. Because we can't be asking. Normally, we can't expect students to be. I'm listening the class only for 5 minutes. I will just ask them or maybe point out them. And then? If if I feel they are not. Still understanding the concept. I might, uh, call out them and teach them. Personally, maybe like a batch of 4-5 students. If needed. By probably by the. 1st 2 weeks. All three weeks, we know how these students are. How they perform?

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

So how was my performance?

I can't give detailed feedback during the interview, but I appreciate you engaging with the questions and sharing your experiences. Anything else you'd like to ask about the role or expectations?

Yeah. Can you elaborate about the role? How was the work life in VIT? Can you just elaborate bit on it?

Of course. The Assistant Professor role at VIT focuses on teaching, guiding student research, and supporting curriculum delivery in emerging technology fields. You'll be expected to lead classroom and lab sessions, mentor students, help with research and publications, and participate in departmental responsibilities. Work life here is collaborative and student-focused, with an emphasis on both academic excellence and practical learning. Anything else you'd like to know?

Projects related to healthcare currently, like some research centers in your organization.