Hi Abjul.

Please start by introducing your professional journey in academia

Good evening, Arjun. I'm Doctor Rajiv. I'm currently working as Assistant Professor senior grade. At Vellore. Sorry, Welltech Rangarajan, Dr. Shagundala Institute of Science and Technology.

I have completed my research in the year 2024. From Vellore Institute of Technology. My research focus on. Optimizing focused web crawler using machine learning technologies.

Alright, Dr. Rajiv, since you've been working as an Assistant Professor, I'd like to hear about your classroom environment. What's one foundational concept or topic you particularly enjoy teaching, and how do you typically introduce it to your students?

Okay, so your research was on optimizing focused web crawlers with machine learning. That's a pretty technical area. When you introduce the basics of web crawling or machine learning concepts to your students, what's your go-to strategy for making those ideas stick, especially for students who might not have a strong background?

After that. I have joined as assistant professor senior grade. At Welltech University. I have over 10 years of teaching experience from various colleges. I have started my career as a lecturer at Saint Peter's Engineering College. And then I have. Worked as assistant professor at Indira Institute of Engineering and Technology. Hande in Hive, joined as assistant professor at Welltech University. After that. I pursue my. Research at Vellore Institute of Technology as a full time research scholar. After completing my research, I have joined at. Again, I have joined it. Welltech Ranjirajan, Dr. Shagunthal. R&D, Institute of Science and Technology. At Abadi this is mail. Career road map. I have published my research paper in reputed journals like. Expert systems.

And others. My research. Purely focus on.

Alright, so you've got over a decade of teaching experience across several institutions, and you've published in journals like Expert Systems. When you're guiding students through their first research project or thesis, what's one challenge you see most often, and how do you help them work through it?

Okay, so your research focus is on optimizing focused web crawlers using machine learning, and you've got a long track record across different institutions. When you're teaching students about web crawling or introducing machine learning for the first time, what's one foundational concept you always emphasize in your classroom?

Developing focused Webrawler. For crawling. Events.

That is my research. I have handled various subjects like Artificial intelligence, Machine learning, Software engineering. Computer graphics.

Alright, so your research is centered on developing focused web crawlers specifically for events. When you're teaching students about the fundamentals of web crawling or information retrieval, what's one concept you find students struggle with, and how do you help them grasp it?

This is my small introduction. Ar.

Okay, so you’ve taught subjects like artificial intelligence, machine learning, software engineering, and computer graphics. When you’re teaching a topic like machine learning to undergraduates, what’s one misconception or stumbling block students often face, and how do you help them get past it?

Alright, you’ve handled subjects like artificial intelligence, machine learning, software engineering, and computer graphics — that’s a solid range. When you teach a topic like machine learning to undergrads, what’s the first core idea you make sure they understand before moving into algorithms?

The core idea about machine learning is.

We have huge amount of data surrounded by. Us so we need to utilize the data. So instead of. Giving input. And finding. Making logical algorithm. On that input and producing output. What we are doing here? But you think this? Data, huge availability of data. We are you making use of the data? We are finding the logic from the data. That is the insight we need to 1st. Deliver to the user. So. We need to make use of that data. By using the data. We need to get some insights or. Patterns from the data without any explicit programming. So we need. We don't want to give logic. Instead of giving logic. We need to give input and output from that. Input and output. We need to. Produce logic. This is the simple thing. We need to. Give to the students. While we are starting the machine learning subject.

You started to say the core idea about machine learning is— but didn't finish. Can you walk me through exactly how you explain that foundational concept to a student seeing it for the first time?

Alright, so you emphasize to students that machine learning is about letting the data reveal the patterns, rather than coding the logic explicitly — making that shift from traditional programming to data-driven learning. When you're introducing this idea, do you use a specific example or classroom demonstration to help them grasp it right away?

Suppose.

We need to find bin. Yeah. Suppose. We need to find the price of the apartment. So. Instead of explicit explicitly writing the. Code to find the price of the apartment. What we are go doing in machine learning is. We have huge amount of apartments. And their price data. So what we are going to do, we are giving these data. As inputs to the machine learning model. So machine learning model, what it will do? It will learn from the data. So for. Which square feet area? Which amount? Will be. The price for the particular. Uh, apartment. And which area, whether it is rural area or urban area based on the area? The price will vary. So it. Identify the insights or. Patterns from the data. From the data. Whenever we are giving new input to the machine learning model. It will predict the price of the particular. Apartment that is the. Basic. Or practical thing I will teach to to the students.

Okay, so you make sure students see the difference between traditional algorithms and learning from data right from the start. When you first introduce this to them, do you use any practical example or classroom demo to help that concept click?

So when you use the apartment price example, you show students how a machine learning model actually learns those patterns and uses them for prediction. How do you get students actively involved in understanding this idea — do you use any hands-on exercise or a demo in your first class?

Yeah, surely. I will do some demo. In my first class.

So I will separate.

The students into two groups. The first. Group A. Uh, apartment ward. Apartment area. The area of these are square feet. Of the. And the second group is the price of the apartment. So whenever. Based on the X axis. Consider. The area will be X axis. The amount will be Y axis. So whenever the area increase. The amount will also increase. Based on that. I will give some pack. Typical demo to the students. So that. The students can easily understand the concept. Easy logically.

You mentioned your paper in Pattern Recognition Letters on keyword weight optimization in event-focused web crawling. How did you translate the findings from that research into something your undergraduate students could actually grasp or apply in a classroom setting?

Alright, Dr. Rajiv, let’s discuss your research productivity from an institutional perspective. Given your experience optimizing web crawlers with machine learning, where do you see the most promising and fundable research directions in this area over the next three years? Which specific funding agencies or grants would you target to support this work?

So while I am talking about. Focused crawler. I need to crawl. The medical relevant web pages. Sunday Web. So whenever I need to focus on specific disease. And the symptoms of the specific disease. And we need to suggest some medicines for the specific disease. We need to crawl all the relevant information about the disease. About the. Medicines and about the symptoms. So. I can use. Focus crawler for retrieving all the relevant web pages. On that particular disease, on that particular. Symptoms on that particular medicine. So that. This research work. Will mainly focus on funded agencies like medical. Agencies. Medical related agencies. Purely interested in these types of project.

Medical-based focus crawl. While I am working with medical based focused crawler.

So you're aiming to apply your focused crawler work to the medical domain, targeting agencies interested in healthcare data and disease information extraction. In terms of building a sustainable research group around this, what specific interdisciplinary collaborations would you pursue within the institute to strengthen your proposals and overall research impact?

So you’re positioning your focused crawler work for medical information extraction, aiming to interest healthcare agencies. If you were tasked with building an interdisciplinary research group around this, which other departments or expertise would you bring in to strengthen both your research outcomes and your funding proposals?

Yeah, surely biomedical department.

So you’d collaborate directly with biomedical experts to make your focused crawler more impactful in the medical domain. Let’s shift for a moment — you’ve mentioned guiding student projects before. Can you share one student project you mentored that connected to emerging technologies, and explain how you ensured it reached a successful outcome?

Medicine are the symptoms. And the those things they will clearly identify. So I need to. Interact with biomedical departments so that the work will be very easy to. Complete.

Okay, so you’d have the biomedical team identify the medical details and you’d focus your crawler on their targets. Before we move forward, can you share an example of a student project you’ve guided that tied into emerging technologies, and what you did to ensure the student saw it through to completion?

Yeah, I have guided many projects. In that project, one important project is. In those project, one important project is.

Sounds like you’ve guided several student projects — you were about to describe one in particular. Go ahead and walk me through that project and how you helped the student succeed. You were mentioning a particular project you guided—can you describe what the student built and how you made sure they finished it successfully?

The.

Makes sense. Now, let’s shift to your approach in the classroom. You mentioned using hands-on demos in machine learning—suppose you had to teach a theory-heavy topic to a large class but weren’t allowed to use slides or traditional lectures. How would you design that session so students actually engage with and wrestle with the material themselves?

LLM, large language model. Which is used. For identifying the higher studies. Of the rural students. So rural students generally not aware of. After completing their 12th standard, they don't aware of which. Uh, IS studies they will. Focus on. Which higher study they need to? Hand take care of. But based on these LLM. If the, LL if the. Prompt they have given is uh. What are all the interest subjects they are interested in? What is the 12th mark they have secured? Those things. We have given us. Prompt our our input. The LLM generate. The related scores uh courses based on their interest particular interests. So that the rural students are. Benefit with this kind of large language.

Makes sense. Now, beyond teaching and research, what role do you see yourself playing in department-level governance — things like curriculum committees, program reviews, or accreditation efforts?

Number of subjects. So they will. Create a contents in that particular subject. So after doing. Those videos see content videos. I am the coordinator to collect all the E content videos. And check the quality of the econtent video once. The video is.

Well versed are qualified 1. We need to post. It in uh, YouTube channel. So that students can. Benefit with these kind of econtent students. So this. These are all the. Role I have done in my. Currently working in university.

Alright, so you coordinated the collection and quality check of e-content videos for accreditation — that’s a concrete example of institutional stewardship. When you were ensuring quality, what specific criteria or standards did you use to decide whether a video was acceptable for submission during the NBA process?

Alright, so you coordinated the collection and quality check of e-content videos for accreditation — that’s a concrete example of institutional stewardship. When you were ensuring quality, what specific criteria or standards did you use to decide whether a video was acceptable for official use?

The first thing is. The duration of the video. The duration. Should not. More than 15 minutes because nowadays, students. Are not interested to watch long videos. So each topic should cover within 15 minutes. If it is a mathematical concept. Then only we can allow more than 15 minutes. And the quality of the. Video should be high. In terms of resolution. And audio quality we are. Uh, checking properly? If it is satisfied with time, duration, video quality and audio quality. We are accepting that video. And posting in. YouTube channel.

Of our institution. Official YouTube channel.

Alright, you emphasized video length, resolution, and audio clarity as your main standards. What would you do if a faculty member submitted a technically excellent video that was twenty minutes long, but the topic genuinely couldn’t be broken down further without losing context?

Alright, you emphasized video length, resolution, and audio clarity as your main standards. What would you do if a faculty member submitted a technically excellent video that was twenty minutes long, but the content was so thorough you felt it would genuinely help students—would you still enforce the time limit, or make an exception?

If it is a really. Good video which contains important topics. Surely we can accept the video. Because at last. The students will be. Beneficiary. So we are concentrating on student beneficiary. If the uh video is really quali, uh, qualified 1. And really the. Content is good. We are accepting that video. Even if it is 20 minutes.

Understood. Now, imagine this scenario: a student files a formal complaint alleging your grading was biased, but at the same time your department head is putting pressure on you to improve pass rates. How would you navigate the tension between maintaining academic integrity and handling these institutional pressures?

Getting bias? I never go with grading bias. Even though my superior.

Is pressure putting any pressure on me? I never allow. Any grading by us because grading is purely. It depends.

You’re clear that you’d stick to fair grading even under pressure. But if the department head disagreed with your decision and insisted you change the grades, how would you respond?

Alright, so even if your department head is pressuring you, you’re saying you’d never compromise on fair grading. But let’s say your department head outright disagrees with your stance and insists that you change the grades—how would you handle that situation?

Students Life. So I never compromise. My grading style. Whenever these. Superior. Put any pressure on me. So the real thing is. I will give my. Original marks whatever the students secured in the. Paper or lab? I will give the proper marks. Even though. The. Superior give. Any pressure on me? I never worried about that pressure because. The grading is purely depends. On students like. So I don't want to confuse students life with the. Uh, superior pressure.

Alright, shifting gears. Given your background working with large language models and biomedical applications, do you have any industry collaborations or connections that could provide internship or placement opportunities for your students?

That kind of support. And now?

Uh, because I have joined uh. Before 2 years.

Okay, so you haven't had the chance to build those industry connections yet. If you were to join VIT, what concrete steps would you take to start building collaborations with industry or bringing practical exposure into your classroom?

So I am identifying the students who are all. Very interested in. LLM. I'm developing LLM. And also in biomedical topics. So I didn't find any students till now. If I found. The really interested students. Surely I will go with. These types of consultancies.

Okay, so you joined your current institute just two years ago and haven't built industry partnerships yet. If you were to join VIT, what concrete steps would you take to start developing collaborations with industry or bringing real-world problems into your classroom?

Okay, so you've only been at your current institution for two years and haven't found students interested in large language models or biomedical collaborations just yet. If you were to join VIT, what concrete steps would you take to start building partnerships with industry or bringing real-world problems into your classroom?

If I joined in industry. IT industry.

You mentioned that if you joined an IT industry, you’d pursue those types of partnerships. Since VIT emphasizes industry-academia collaboration, can you share how your experience with consultancy or industry-adjacent projects could help create impactful opportunities for students and research here?

When I. First thing. I need to find the research gap.

Between the existing system. And the current system. So if I found gap. Then only. I can do some research on that particular work. So I will. Discuss with my clients. What are all the expectations? Our research gap? They are. Finding SO. Once I clearly find a research gap. I can. Work 2. Build the gap for to satisfy the gap so that. That will be the challenge for me. So I need to find the problem first. Research problem. And then I need to. Select some students, those who those are all interested in this research. Once I found the. Interested students. I can explain the concept. What is the work we need to do? So I can get the work from that them. After that.

So, you said the first step would be identifying the research gap. Can you walk me through a specific example from your past where you did that—how did you pinpoint the gap and what was your next step in forming a collaboration or project around it?

So you said the first thing is identifying the research gap with clients, then selecting interested students and explaining the concept before starting the work. Can you give me a concrete example from your past where you actually did this—maybe a project where you worked with an external partner or client and involved your students?

Give you completed the prototype.

I will give the give to the company peoples. Once they.

So you’d only move forward with building a prototype after getting buy-in from interested students and clarifying the research gap with the client. Can you give me a concrete example—maybe from your last two years—where you actually identified a gap, guided a student project, and saw it through to a deliverable, even if it wasn’t industry-funded?

Find found. The prototype is good. Or they will give some feedback. Based on the feedback, we will do some rework. On that particular prototype. This will be iterative iterative process once the iterative process is completed. We will get the. Complete product. If we satisfy and the company people satisfy. If I uh. Get the win win situation. Then we will deploy the product.

Right, you’ve talked a bit about building prototypes and collaborating with companies once there’s a clear research gap. Switching gears, let’s dig into some core skills for this role. Can you walk me through a time when you applied multimedia or artificial intelligence tools to solve a real-world media problem?

So you described working iteratively with the client, building a prototype, collecting feedback, and involving students in the process. Can you give one concrete example of a research gap you identified in your recent work and how you explained it to students to get them engaged?

Yeah, sure.

So here we have number of large language models. Nowadays.

Okay, you've described involving students in research and building prototypes iteratively with industry feedback. Shifting topics, how do you structure your theory and lab courses to ensure students with different learning speeds all achieve the core outcomes by the end of the semester?

Alright, you started talking about large language models. Can you spell out a specific gap you noticed in the current large language model landscape, and describe how you’d communicate that gap to a group of master’s students to get them genuinely interested in tackling it?

But whenever we are giving any prompt. They give answer. But the answer contains lots of.

Hallucination. So hallucination is the big research gap. We're finding nowadays, so we need to identify the gap first. So Hallish Nation is the big research gap we have. We need to focus on recover that hallucination. We need to reduce that hallucination. So how to reduce that hallucination based on various? Techniques. So that is the research gap. I have found a number of large language models. When? We have found the research gap properly. We need, we need to focus. On reducing the research gap or satisfying the research gap. So hallucination is the major research gap I found in LLM. So we need to do work. To reduce the hallucination. That is the main research gap I have found. In recent LLMS.

Okay, you were saying that large language models give answers but the answers contain lots of— go on, what’s the specific issue you’re highlighting with their outputs?

So you see hallucination in large language models as a core research gap, and you're focused on finding techniques to reduce it. When you're teaching about this issue, how do you structure your theory and lab classes to help students really understand and address hallucination in LLMs?

Hallucination first. I need to.

Clarify the students what is hallucination? OK.

Okay, you've thought a lot about research gaps and getting students involved. Shifting topics, tell me about a time you had to evaluate students’ performance—what approach did you use to ensure the exam or assessment was fair and actually measured what they learned?

Alright, so you'd start by clarifying to students what hallucination means in the context of large language models. When you explain this concept in a classroom, how do you structure the discussion and make sure students really grasp both the technical meaning and the practical implications?

When the student. Question 2. Any LLR. That is, he's giving. I am very tensed. What I need to do? That LLM replies. Go and. Kill yourself. So this is the very bad answer the LLM gives to the particular student. So this is called hallucination. So we need to reduce the hallucination. If the students are using LLM. They will. Get some these types of wrong answers. So this should be reduced. So this is called halogenation first. In theoretical way. I explained what is hallucination.

Alright, so you make sure students understand what hallucination means by giving them examples with LLMs giving harmful or nonsensical advice. Shifting topics for a moment — walk me through a specific time when you were responsible for evaluating students’ performance, like during an exam or project, and explain how you ensured your evaluation was fair and objective.

Once hallucination. Is clear to the students. I will go to some demo classes. Where vectorspace model plays an important role. So we need to. Identify the exact meaning of the word, that is semantic meaning of the word. Won't the semantic meaning is clearly a? Uh, identified. We can easily reduce the hallucination. So for that, uh, vector space model. I can. Deliver What is TFIF? What is vector space model? What are all the various uh. Meaning technology word meaning technology we can identify. Once I clearly explain the students. In this lab demo. They can.

Use some practical exercise so that they can implement the. Large language model easily.

Alright, so after you’ve explained hallucination and shown how vector space models and semantic meaning tie in, you run demo labs to make it hands-on for students. Switching gears for a moment—can you tell me about a time you were responsible for evaluating students, like during exams or project reviews? What was your process to make sure your grading was fair and consistent?

Alright, so after your students get hands-on with implementing large language models and see how techniques like vector space models help reduce hallucinations, they’re likely to encounter unexpected results in the lab. When a student's implementation doesn't work as intended, how do you evaluate their understanding and performance in such a scenario?

So I will evaluate the students understanding first. Based on asking questions. What is Halation? What is LLM? What is prime? These types of basic questions first I will ask. Once they are understand. Then only. I can. Go death, that is. How to implement the? By using various techniques. So this is the evaluation method I used. On my toilets.

You mentioned you often start student evaluation by asking basic conceptual questions before moving on to practical implementation. When you’re actually grading lab work or exams, how do you ensure your assessments are fair and consistent across students?

In lab. I will give the different scenario. For different students. So what? I have to do is. I have to. Separate the entire students into number of small groups, that is clusters. And I will give. Each scenario for. Each groups. So. They need to understand the logic. First, once they understand the logic. I will give some syntax and. Uh, syntax and. Keywords. What are all the keywords we need to you to, uh, use? What are all the syntax we need? Everything I will teach clearly. Once they have given me the logic and syntax and keywords. They will start their implementation. So once they have start their implementation. I will ask. Questions to them. I will check. Uh, regularly. They are implementing properly or not. If there are any errors arise. I will rectify those errors. So in this kind of. Manner I will evaluate the students properly.

So you break your students into clusters and give each group a different problem scenario, then check in regularly as they implement and address errors together. When you’re running exams or handling university-level evaluation duties outside the classroom—like invigilation, question paper setting, or centralized grading—what steps do you take to ensure the process is transparent and unbiased?

Whenever I have en engaged with other duties. Like university exams? Or any other?

Okay, so when you're assigned to university exam duties, or other centralized evaluation tasks, how do you handle situations where students or other faculty question the fairness of your grading or your procedures?

Surely I will make sure that. Any ill mother studies online quizzes I can conduct? So that. Students can actively involve. In that particular exercise. After completing my work I go. Check whether they have done properly or not. So by using ALM methods like online quizzes. And. Uploading some UH PPTS to the students. So that they can clearly. Analyze the PPT and understand the PPT. Once they understand the PPT. I can uh. Give some. Test online, test to them so that they can. Write whatever they. Understand. So these types of methods I can use. To effectively engage the students. In my. Other works. Whenever I have some other words.

Alright, so when you’re pulled away for other duties, you use online quizzes and PPTs to keep students engaged and check their understanding later. Thinking about your overall teaching style, how do you make sure your lectures or labs stay structured and clear for all students, especially those who might struggle with the material?

First thing.

I will identify. The fast learners and slow learners. The fast learners are the students. Who can understand easily whenever we have shared any PDF materials? Are demos. So. I can partition them.

So you start by making sure your students understand the basics and then move into deeper material, using online tools and resources when you’re pulled away for exam duties. Now, for this role, communication is critical—when you’re explaining a complex topic like vector space models or LLMs to a class with mixed backgrounds, how do you break it down so everyone can follow, not just your top students?

So you split the class into fast learners and slow learners based on how quickly they pick up the material. Once you've identified those groups, how do you adjust your teaching or communication style to make sure both groups actually benefit and nobody gets left behind?

I will identify the slow learners. So I will contact them personally. What are all the difficulties? Do you have? They have in understanding the logics. Are concepts, so if I found the difficulties. It is easy to explain them in a proper way. Their understandable way that is the. Thing I will. Do.

So you reach out personally to slow learners and find out what’s tripping them up, then tailor your explanations to their needs. When you’re teaching a complex topic to a group with mixed abilities, how do you keep your class on track without leaving either the fast learners or the slow learners behind?

Yeah, whenever I'm handling complex topics. It is very difficult to understand. I will start with some real time example. So whenever we give some real time examples. Then only the. Can come to the real time so that they can understand easily. Whenever. We get the attention of the students. We can easily share the concept. To the students. So how to get the attention of the students? That is the main question. By using real time examples. Is the only way? We can get the attention of the students. So whenever I start the complex concept, concept. Like complex techniques, surely I will start with some real time examples. So that. Students can easily understand the concept. Without any bias even. They are so learners are faster.

Starting with real-life examples to hook student attention makes sense, especially for tough material. Switching gears, Professor, to your research publications—can you walk me through a recent paper of yours, and explain what set it apart from others in your field?

Recently I have published. A focused crawler based research paper. Which is used stochastic gradient descent. So here. Instead of. Crawling all the web content. I have used focused crawler. Which covers only relevant web pages in the particular topic. That is called focus. Here the keywords plays a major role. To find the relevant web pages. So how to? Find the important keywords or how to. Assign the weights of the keyword that is the main. Research problem. So here I have used stochastic gradient descent optimization. Which is. Used to give the prop. Per optimized weights to each keyword based on their importance. So that. We can, clearly. Get relevant web pages. From the web. So this is the research work I have done.

This is the novel work I have done. That is, instead of. Getting all the information. I have select only relevant web pages. Based on the optimization called. Stochastic gradient descent. SGD.

Alright, so you tackled the core problem of keyword weighting in focused crawlers by applying stochastic gradient descent—using optimization to surface the most relevant web pages. When you were preparing that paper for publication, what steps did you take to ensure its novelty and that it met the standards of reputed journals?

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

I didn't get you.

You clarified your use of stochastic gradient descent to optimize keyword weighting in your focused crawler work, which is a solid technical contribution. We're all set to wrap up, Professor. Is there anything you'd like to ask about the role or the team before we conclude?

The session was good. It is very interactive. I'm really enjoying the session. That is what I want to say.

Glad you found the session engaging, Professor. Before we finish up, do you have any questions about the role or the team?

Yeah, if I got selected. Based on my experience. Which designation I will get? Do you have any idea?

Based on your experience and qualifications, Professor, you'd be considered for either the Assistant or Associate Professor designation. The final title usually depends on your publication record, years of teaching and research, and how those align with institutional criteria. Anything else you'd like to know before we wrap up?

Nothing. Nothing. That's all.