Yeah, myself, doctor Vijay Marelpa.

I'm having Vijay Kumar helpful.

Thank you for sharing, Professor. Please start by introducing your professional journey in academia.

Thank you for that, Professor. Which foundational topic in your discipline do you find most engaging to introduce to students in their very first semester, and why does it resonate with you?

C programming and data structures. For the first year students.

Yeah, sure, why not? In the programming concept, generally the student cannot understand how it is executing actually. So background because we will write the set of code couple of lines maybe 10 or 20 or whatever it may be and when we are going to the execution part. The student he doesn't know what is happening, if it is a loop or something, how it is executing each line by line. So such kind of things we can make animation. As well as the interactive. Uh, in the digital content of e-learning, uh, using the e-learning, uh, softwares like storyline and etcetera, things, we can record the screen recording and we can make some interactive design as well as the simulations so that the student can interact here. Like example any. Uh, stack or something? Uh, stack or queue? When we are using this kind of data structure. I'm developing AC program. So how it is executing? How? Well, they're pushing out. An element in the stack. And how we are. Uh, going to the hopping that. So that we can make some kind of animations. Uh, while it is executing the program. We can split the screen into the program and as well as the second. Off part visualization part. So in element is going to. To be, uh, pushing like that one, it is entering into that some kind of box. And so like that one we can. Uh, show uh to the student. And along with that one, we can also. Uh, display the. Um, random, some memory. Numbers or memory locations. So with that. Some when it is. Pushing the memory, how it is increasing the next level? Uh, if based on the. It may be the integer or uh. Character type so we can go the. The like the. So this kind of showing. Visualization to the student, he can easily understand and when. Similarly when we are popping that one top of the stack how it is? Uh, removing the element or how we are we are picking that element. So that also after winning that one, the from that. Mocks. We can, uh. Remove that element. We can make that one as the empty so he can visualize that one. So how it is? Uh, happening what kind of operations we can perform. So that is very easy to understand when we are going with the visualization and as well as the interaction. So in the interactive side he can we can make some. Animations like clicking the element, dropping or dragging something. Like that one we can do here.

Given your experience developing digital content for C programming and data structures, can you describe a specific example of an interactive or multimedia tool you created to help students grasp a difficult concept in these courses?

You mentioned using tools like Storyline to create interactive simulations for stack operations—can you walk me through a specific challenge you faced when designing these visualizations for students who struggled with abstract thinking?

Call here for example if you go for in depth lead data structure side. The programming is more complex here. The length of the programming is more and more. Uh uh, lengthiest 1. So there it is. Tough to present everything on a single screen.

So simulation site. Because, uh. It may requires. Uh, multiple, uh. Uh, functions. It is involving with that one. So that is a challenge here.

Find a visualization is a little bit tough. We can, but we can provide all those things.

Understood. In situations where the simulation became too complex for a single screen, what specific strategies did you use to break down or sequence the visualization so students could still follow the full process without getting overwhelmed?

Understood. Shifting focus, Dr., VIT places strong emphasis on interdisciplinary research and externally funded projects—can you share a concrete example from your career where you collaborated across departments or secured funding, and how you navigated the challenges involved?

Functions. Uh, cross functional teams I have work. Where? Here I can do. Uh, walking my research reinforcement learning with the human feedback side, umm, I realized that that the purely technical optimization was not sufficient. Especially for a safety critical application. So I initiated A collaboration with the. Uh, other colleagues like uh, psychology and as well as the human factors domain aspects. So to better understand. How human feedback can be. Modeled, uh. Reliably. So this is like this one I can. Worked with the. Cross functional. Departments here.

Let's turn to teaching methods. Suppose you are assigned a large introductory course in your field and are not allowed to use traditional slide lectures. How would you implement an active learning strategy, such as a flipped classroom, to engage students and ensure deep understanding?

Great, earlier I worked for the.

Teaching almost 10 years. So if I'm not allowed to use a traditional lectures, I would adapt the. Flip the classroom, uh. Combined with active learning techniques. To injure the both engagement and as well as the deep understanding.

Let's discuss your experience with institutional stewardship. Suppose your department needs to increase its average research output per faculty member to meet a new accreditation benchmark. What systematic steps would you take to improve this metric while maintaining academic integrity?

Professor, imagine a scenario where a student alleges grading bias, but the administration is pressuring you to adjust grades in order to meet a mandated pass percentage. How would you approach resolving this conflict while maintaining your ethical and professional standards?

So it requires. Some preparation phase. Uh, before the class and during the class active living phases. Like a class time would be. Used entirely for the active engagement. So. Payer sharing. Uh, live problem solving and repeat interaction with the. Among the students. Find some if it requires some coding activities. Handling large classes for the. I would divide the students into smaller groups and assign the roles like like a role player or leader or code or something regular to win. Ensure all the participants. Perspiration. And we can use the kick polls pieces. Monitor via teaching. As a sense if um.

Possible and also. After the class. Uh, like a coding assignments? Conceptual quizzes. Discussion now. Forms and assessment strategy I can use like a weekly puzzles and practical coding task. And as well as the conceptual problem solving. So this all comes at best outcome I can say.

You mentioned dividing students into smaller groups with assigned roles—can you give a specific example of an activity or problem you used in this format, and how you measured whether it improved student understanding?

Thank you for outlining your approach. Now, Professor, imagine a scenario where a student alleges grading bias, but the administration is pressuring you to adjust grades in order to meet a mandated pass percentage. How would you approach resolving this conflict while maintaining academic integrity?

This is it happened, you know, not only the academic, even it I faced such kind of things in the. Even industry side also. So I would approach this situation with a. Strong commitment to fairness. Transparency. And academic integrity. While also maintaining, uh, constructive. Dialogues with uh. In the administration. So first of all, I will address the students concerns first. So I would, uh, first. Uh carefully review the students evaluation to ensure. There is no unintentional bias or any error, so recheck the. Grading rubric. Comparing with the. Other students work. And I enjoyed the transparency. I would clearly explain the. Grading criteria and provide evidence based justification for the. Uh, assigned marks.

Let's discuss industry alignment. How would you leverage your professional network to create internship or employment opportunities for students, particularly in emerging technology fields?

So here I handle the administrative pressure like. Regarding the. Pressure to adjust. Greatest to meet a. Pass percentage. I would respectively communicate that. Grades should reflect actual student performance and uh. Learning outcomes.

And also offer a constructive alternative. I will find out. So instead of adjusting the rates I.

Now, I'd like to verify a few specific qualifications required for this role. Can you describe your direct experience conducting student evaluations and exam duties, detailing the methods or systems you have used to ensure objectivity?

Right. Can you give a detailed example of an exam or assessment you designed for first-year programming or data structures, and explain how you ensured its fairness and reliability?

For the role assistant or associate professor. Uh, essential, uh. Qualifications, typically. Uh, at. Doctoral degree with some best experience where I'm having a academic and as well as the industry both experience. I have. And. Come to the student evaluation side. I have been actively involved in the designing and conducting the student evaluations, including the assignment. Business lab assessments. And in the semester examinations including the designing the question paper and evaluation side. So here it is how I will do means. The Rubik based evaluation I will go and. Anonymous grading if possible. Blind. I understand that answers keys I will prepare. And moderation process I cross check. Borderline cases ensure the uniform. Grading across all the students. And automation tools if. Uh, university allows. I will use the automation tools. For the ability.

So come to the. A best detailed example. As you said, a conceptual MCQ. Uh, for the structure of the exam. Uh, any dark structures have you used? Uh. Find uh. Focusing on for example, like LIFO. Find a complexity. So in the next section I will go with. Dry run problems like uh. Uh, trace stack operations step by step. Next to I will ask the coding questions, implement the stack using some arrays or. Uh, any other links like that? Then section next level I will go with some. Application problem. So balance through the parents design, the post fix evaluation, such kind of things are. So here in the all the area I will go with some kind of bloom stacks anime also I will apply. So. We can go with the easy media mind. All the cattle. So this. Kind of things I can. Uh, use all those areas.

You mentioned using automation tools for evaluation when permitted—can you specify which tools you've used and how they impacted the efficiency or accuracy of your grading process?

You mentioned moderation and the use of Bloom’s taxonomy in your assessments—can you walk me through a situation where you had to defend your grading or exam design to an academic review board or external panel?

Yeah, I I was a. Working in the.

Understood. You referenced using blind grading and answer keys as part of your evaluation process—can you give a detailed example of how you handled a borderline case, and what criteria tipped your final decision?

And analytical. Options rather than. Purely theory ones. So after evaluation, there was a concern from the panel. That the paper might have been slightly challenging and as well as the for the average students. So here the challenge was, the concern was mainly. About the fairness and whether the exam aligned appropriately with the expected difficulty levels for the first year students or not. So my justification is here. I explained my design using the bloop taxonomy. And as well as the outcome based evaluation principles.

So when you defended your exam design to the panel, what specific data or examples did you present to demonstrate that your assessment was balanced and aligned with learning outcomes?

So the evidence here I presented was a detailed question. To outcome mapping. And the clear making rubic. Sample student responses across the. Performance levels. So this all I have done. So after discussion. With the panel, I agreed that assessment and they, they agreed actually assessment was, uh, academically sound, but uh, uh, they suggested, uh. Adjusting the weight is uh. To what the? Foundational questions in the. Future iterations.

Whichever I took positively and I have incorporated such kind of things.

Thank you for clarifying. You mentioned moderating grading and using detailed rubrics—can you share an instance where automation tools directly impacted the accuracy or efficiency of your evaluation process, and specify which tool you used?

Thanks, Professor. Now, can you describe your approach to guiding student research or final year projects, specifically how you help students narrow down a broad topic into a focused, feasible research question?

I used to guide for the students in the research to move systematically from a broad area to focus. Hezbollah as well as the impact research problem. While ensuring the student develops independent. So I start the board. Then structure like initially students come with. The broad topics like uh, example, uh yeah in the healthcare or machine learning applications, et cetera. So I ask them to read to. Uh, couple of uh. Survey papers like at least 2. UH-32. Five papers like that one. And then identify the sub areas. So then next. We will find out the identifying the interest and as well as the feasibility. And as well as the available resources. Data sets, so time constraints. So etcetera, I will go with all those things. Then I use a funnel approach like from domain problem to. Specific questions. So like. Yeah, in healthcare. So I'm narrowing that one the. Predicting diseases. So like that one, so we can focus on something. So a good research question. Uh, like a specific point as well as the measurable? And also solve it. So I will explain such kind of things to students. Then I will go with the defining the clear objectives, breaking into the milestones like a literature review, data set collecting in the. Uh, model implementation and how we will evaluate? So this is continuous. So I will continuous monitoring all the students, how they are doing everything. Then we'll go in with the handling with the weak students. Uh, for UH students who. Uh, struggle. A slightly reduce the complexity, but. Maintain the levering objectives and as well as the ensuring they still complete and meaningful. Uh, projects. So this way I will do.

You emphasized continuous monitoring and adjusting project scope for struggling students—can you provide a concrete example where you helped a student reframe an overly ambitious project into something achievable, and what criteria you used to guide that process?

Yeah, sure. Actually I am a PMP certified. Industrial. So. So I encountered this with the. Pioneer students earlier also.

You mentioned your PMP certification and prior experience—can you describe a specific student project where you applied project management techniques to help the student create a clearer timeline and actionable milestones?

So here the. Problem was uh. Investing, but clearly. Two ambiguous sometimes given constraints like a limited, uh, computational resources. Or lack of real world data. And sometimes the semester time deadlines. So here. I worked on the with the students reframing the project. In such areas. So here the broke the system into the components identifying 1 core problem to solve like that. So and. Instead of full autonomous driving. So we focused on implementing a. Uh, reinforcement learning isn't like, for example, uh. A land keeping in a. Simulated environment. So like that one? So this is for example. So we will go like uh. Going with the. Criteria feasibility, So. Can it be completed within the? Deadline 1 semester. Amit the available tools. Uh, clarity will uh will define the input sign outputs. As well as the evaluation metrics are available. Uh, like a lane derivation reward functions? And as well as uh. Uh, levelling values. Should still cover. On the core concepts. Reinforcement learning basics policy learnings evaluation side. Uh, and also execution supports. My further I will divided the project into the milestones like literature reviews and as well as the environmental setups and. Uh, basic soft. IRL implementations performance evaluations. So like that one I will go all this. Uh, criteria.

Understood. You mentioned defining clear inputs, outputs, and evaluation metrics for projects—can you give an example of how you helped a student select or design a dataset when real-world data wasn't available?

Practically as now. We have collected most of the data.

Understood. When you needed to generate synthetic data for a student project, what specific tools or methods did you use to make the dataset realistic and relevant to the research question?

From the. Free like and all the resources. But, uh, sometimes, uh, it may not, uh. Possible so I graduate the student to design A simulated data set. Uh, our environment. That could approximate real-world conditions while. Remaining manageable. So we followed the like a structured approach like defining the input sign outputs where the input. Uh, state variables like positions. Environmental conditions and as well as the. Like output rewards or some kind of labels. I know we using the simulation tools. Like nowadays we can use open AI. Uh, like that one we can use. And the synthetic, uh, but uh, we can go with some kind of realistic. Uh, scenarios? Find reputable experiments. And incorporating the variability like to avoid. Uh, over a fitting. Uh, who introduced some kind of variability like uh, noise in input sign as well as the multiple scenarios edge cases? And also.

Got it. When introducing variability and edge cases into a synthetic dataset, how do you determine which scenarios are most essential for students to include, especially when simulating real-world unpredictability?

Since the real human feedback was limited. So we designed like a proxy like rule based regard. Signals. Find a small manually labeled samples for the validations. Um, and evaluation metrics also we went with some accuracy or success rate, uh. Some provide convergence and safety violation side. And validating the. Data sets. So, uh, part of, uh, after all these things, uh, we got the best, uh. Outcome.

Okay. Earlier you mentioned using automation tools for student evaluation—can you specify which tools or software you have implemented for grading or assessment, and how they impacted your workflow?

Come to the Hawaii, user them in the. Student side means for programming assignments. I designed a problems with the. Predefined some test cases. For students, submit the code and system runs. And then multiple test cases they will check and edge cases so like that one and also. I will go with the efficiency and the consistency. Uh, in red feedback. Uh, I'm focusing the shift. So these are all the things I will do and. So I will find out the limitations plus uh. The. So code quality like a logical clarity and as well as. So overall integrating the automation. Uh, improved. Uh, both scalability and as well as the. Quality of. Assessment I.

Thanks for outlining your approach. What challenges have you encountered when using automated grading platforms like HackerRank or Google Classroom for assessments in media or AI courses, and how did you address them?

Uh, like.

Vit. I can expect. The there is a huge number of students.

Right. Since you mentioned scalability and quality improvement, can you give an example where automated grading produced a misleading result for a student, and how you detected and corrected it?

You mentioned the challenge of large student numbers—how do you ensure your feedback remains personalized and meaningful when using automated assessment tools at scale?

Uh, the challenges, uh. Here I faced like a non detriment pic. Office uh, in the AI models in uh, example like uh. Outputs are often per. Probabilistic. Um so 2 correct models. May produce the slightly. Different results.

So here. Instead of exact output matching, I use some kind of threshold based evaluation matrices.

I see. When students’ models produce slightly different but correct outputs, how do you design your grading rubrics or assessment criteria to fairly evaluate their work?

Understood. Can you walk me through how you set your threshold criteria for evaluating AI model outputs, and what factors you consider to ensure fairness across different student submissions?

Where like such as accuracy, F1 score signed as well as a loss values. So here we define acceptable performance range allowing uh. Uh, flexibility in the results. So here the challenges are here like evaluating the code and the quality side, creativity side. So automation focus on correctness but. Uh, cannot fully. Capture so code readability. And the design choices. Innovation. So here we worked with a hybrid evaluation model. So automated grading. So that is for the correctness. And second one is the manual. Yes, so structure, logic and as well as the document. And in the data set side, the challenges are environment dependency. So here, uh. We are providing some standardized data sets. Shared baseline templates. And using some control. Environmental where? Uh, password? So that reducing that reduces variability and as well as the improved. Uh, comparably. Across the submissions. I'm the in the academic integrity side. Automated platforms sometimes cannot fully. Detect the logical. Algorithms. So we use the plagiarism detection tools, uh.

You mentioned using plagiarism detection tools alongside automated grading—can you give a specific instance where these tools identified an issue, and how you handled it with the student?

If. University adding the Viva or shortcut. Differential setting.

I see. When you encounter potential academic integrity violations flagged by these tools, what is your process for verifying and addressing them with the student directly?

I encountered a situation in the programming assignment where automated platform flag would. Two submissions with very high similarity scores.

Once those submissions were flagged, what steps did you take to determine if the similarity was due to legitimate collaboration or actual plagiarism?

I compare the core structures variable naming. Lock follow. Lock flows and as well as the logical flows. Checked time stamps. Submissions. And also looked for the identical mistakes. So this helps me. The another thing is the interacting with the students, uh. Uh, professional handling like that one. So I asked them to explain the approach. Walk me through the. So this agent instrument can explain the logic. Uh, where? A copy? Uh, work struggles to explain that plan. So this kind of, uh. Approach I followed.

Understood. When you asked the students to explain their approach, how did you document the outcome of that discussion, and what actions did you take if the student could not justify their submission?

The thing is here. When we are, uh, evaluating the student side. So I conduct some discussions that I. Uh, the entire process properly documented and as well as the aligned with the instructional guidelines as per the university. So the student details, signed assignment information and the reason for the flag. Summary of the student UH. Find the students explanation. All these things. So additional validations are like, uh. Uh writing a small or related uh snippet signed, modifying the pop-up, some code lively. And explaining the functions, but like that one I do so if the student is not unable to justify for that one. So I'm going to take like a balanced reaction like assigning the. Reduce or 0 marks for the components. Or allowing the resubmission. With some penalty dependency. So this may. Better for the student. In the severe cases I will escalation to the same thing to the head of the department of some of the Dean. Of the university. So this kind of thing I will. Follow.

Thanks for explaining your academic integrity process. Shifting focus, can you describe a time when you guided a student project in AI or multimedia from initial concept through to completion, and what specific strategies you used to keep the project on track?

Using the machine learning. So then we will go for the structured project. So here. Inputs identifying the inputs and as well as the. So inputs are the tweets. And output is the sentimental labels. So evaluation matrices are here accuracy and as precision. Are calling here.

Right, once you helped the student narrow their focus and define their evaluation metrics, how did you guide them through challenges like data cleaning or feature selection during the project?

Then. Baking into the milestones. So, data collections. Data preprocessing. Uh, feature extractions. Model buildings. Evaluation. Finally, uh report and prevention side.

Once you broke the project into milestones, what strategies did you use to keep the student motivated and ensure they met deadlines, especially if they fell behind at any point?

Then we'll I do continuous monitoring. I identifying the. Issues, if anything, are early. Provide the guidance. Without. Giving a. Direct solutions. And I keeping. It on uh. Track regularly. Regular feedback loop sign documentation it. Each page encouraging the experimental. Within the limits. So this all approach definitely provides a best outcome. So this I followed in the event industry. I am aware in the project management professional certification, I learned all these things.

You mentioned applying project management techniques you learned from your certification—can you give a concrete example of how you resolved a conflict or misalignment between a student's proposed approach and the project's goals?

Yeah, I encountered earlier. Such kind of things. Where the student wanted to use a deep learning model. Umm, for relatively. Smaller data set. So the project goal was to build an efficient. And, uh, interpretable. Model for the prediction. But a student was. Very keen on using the complex deep learning models like neural network. So the conflict here was the data set was very smaller and project timeline was very limited. The goal emphasized. Interproductivity. So here I followed the approaches understanding the students perspective. I asked the student to justify why he preferred deep learning. Then next. Data-driven comparisons. Uh, simple model and neural network. I suggested a small experiment. Comparing like that one. Then aligning the project. Goals of. We evaluated the both terms of. Accuracy. Interpretability and resource uses. I focused on guiding the student to. Discover the limitation. Rather than the imposing a decision. Resolution comes like that one based on the results. The student realized that the. Simpler model performed. Comparably, and we were more interpretable. So we refined the approach. To use a simple model. Uh, with proper feature engineering.

That approach shows a strong emphasis on student autonomy. As we approach the end of our discussion, is there a particular research publication of yours in multimedia or AI that you believe has most influenced your teaching, and how do you bring those insights into the classroom?