Good day, Sethurao Gururaja. I'm Arjun. I've had a look through your background and your professional record. I'm interested in hearing more about your work and how you see yourself fitting in here with our faculty. Ready to start?

Please start by introducing your professional journey in academia.

Hello.

I'm unable to hear you.

Hello.

Hello.

Hello, Professor. Please start by introducing your professional journey in academia.

So I think you are not audible to me. I am unable to hear you.

I see. If you are unable to hear me, please check your audio settings or try reconnecting. Let me know if the issue persists.

Do you want me to log out and log in?

If you're still facing issues, logging out and logging back in might help resolve the problem. Let me know once you're reconnected.

Hello, you guys are I'm unable to hear you.

I have enabled all the.

It seems like the audio issue is persisting. If you've checked your settings and everything seems fine on your end, it may help to try refreshing or reconnecting to the platform. Let me know how you'd like to proceed once you're ready.

Hello.

Hello, Professor. Can you hear me now? If so, please introduce your professional journey in academia.

You want me to put my headsets?

Yes, Professor, trying a headset might help resolve the audio issue. Let me know once you're ready.

Hello.

Hello, Professor. Can you hear me now? If so, please introduce your professional journey in academia.

Hello.

It seems there might still be an audio issue. Please let me know if there's anything I can do to assist further.

It's not audible.

I'm unable to hear you.

What should I do?

To whom should I call? I am unable to hear no, no response I'm not getting.

Any sound?

You want me to try in another device?

Yes, Professor, trying another device might help resolve the issue. Let me know once you're ready to continue.

You're not audible.

So I'm trying in my tablet now.

That's perfectly fine, Professor. Take your time to ensure everything is set on your end. Let me know when you're ready, and we can proceed.

So it's not working.

And you are not audible to me. I'm not able to listen anything.

It seems there might be an issue on your end with the audio settings. Could you check whether the volume is up or if your device's audio is functioning properly? Additionally, ensure any connected peripherals, like headphones or speakers, are set correctly. Let me know once you've resolved it, and we can proceed.

Hello.

Hello, Professor. Can you hear me clearly now?

Hello.

Professor, it seems there may still be technical difficulties with the audio. Please confirm if you can hear me clearly.

I cannot assist with that request until you confirm explicitly. Professor, if you'd like to end the interview, please confirm directly with a clear "yes" or let me know otherwise. Thank you!

Hello.

Yes.

Now you are audible to me.

Hello, Professor. Can you hear me now? Please confirm.

Great, I'm glad the issue is resolved. Shall we continue with the discussion?

I think we can start our session.

Perfect. Let’s begin.

Tell me.

Yes, we can proceed into the discussion. There is no doubt for me from my side.

Before we start, Professor, do you have any questions or concerns about the interview process? If not, we can proceed straight into the discussion.

Excellent. Let’s get started, Professor.

Yes, please.

Professor, could you share some insight into your research expertise and how it aligns with the field of computational modeling?

My expertise closely aligns with the computational modeling in different aspects. I'll highlight maybe one or two that is more relevant to my PhD research work and the postdoctoral work that I'm currently doing. So my work closely aligns with the computational modeling through the development and application of advanced ML and AI techniques to model the complex manufacturing process. During my PhD I have created predictive models that simulate dynamic stability and surface form errors in high speed.

Micro milling which are generally computationally intensive problems because it require deep understanding of both physical phenomenon and data-driven methods. Also in addition to that my research work includes numerical methods, sensor fusion and bifurcation analysis. So all of these which are core computational modeling techniques that are used to predict and control the behavior of the manufacturing process. In addition to this as an extension of.

PSD thesis work at last instead of going it on a paper. So I have developed some cloud based platforms and web interfaces for real time process monitoring which demonstrates my ability to translate computational models into practical tools for industrials. Anyhow, this is not an 100% real time problem because we have not implemented in any system because it requires some licensing options and licensing permissions and other. So these can be used as.

Pre machining options where you just enter your machining process parameter and see how much kind of deviations are errors or whether the selected process parameter yields a stable condition or not. So it will be a kind of.

Time reduction methods. So the overall my research bridges theoretical computational modeling and practical AI driven manufacturing solutions and this makes me more suited for the role.

Thank you for providing such a detailed overview, Professor. Your research demonstrates a robust understanding of computational modeling principles combined with practical applications, particularly in manufacturing processes. A couple of follow-up questions: First, could you explain how the integration of AI into these computational models improves their predictive accuracy? Next, for the cloud-based platforms you mentioned, what challenges did you encounter when translating these computational models into practical industrial tools, and how did you address them?

So to answer your first part, so how did I integrate the?

AI for the manufacturing process.

I integrated it by combining the data-driven ML techniques with the traditional physics based models to enhance the prediction accuracy so that the model can be used in a real time processing without any additional training for example. So I developed a multi-sensor fusion deep learning framework that uses real-time sensor data to detect the onset of the chatter vibrations during high speed micro milling.

So this achieved an accuracy of over 93 to 94% and we have tested this on multiple manufacturing process. We trained the model on the micro milling, tested on the turning, then further tested on the macro milling. So in all the cases accuracy was in the range of 90 to 95%. So and for the cloud based method that you asked. So I used the ensemble machine learning methods like the stacking based ensembling to predict the form errors and the prediction errors when compared with the.

Experimental work and they showed less than 10% error. So this improves the precision of the manufacturing process. And this is beyond what a pure analytical models can achieve. Because analytical model are purely dependent on the combination of the cutting tool and the workpiece material chosen. And for every change in the combination the process has to be iterated again and again. But however, as we move to a mill based method.

Reiterating process can be avoided.

Thank you for elucidating, Professor. The way you merged AI-driven techniques with traditional physics models, and achieved such high accuracy across varying manufacturing processes, is indeed impressive. This kind of robustness and adaptability is crucial for real-world applications. Regarding your sensor fusion framework, achieving over 90% accuracy across different setups is quite a leap. Were there specific lessons or findings you encountered during the testing phase when adapting it to macro milling or turning that surprised you? And for the cloud-based ensemble methods, what strategies did you employ to ensure scalability and adaptability across different industrial applications?

So the surprise finding is.

Nothing such because we have kept the input same to the model. The input is itself an image. So in all the irrespective of the process whoever this using we have told them strictly that the model was perfect only when the input is same and the output is same. So you heard the output is a classification problem where we are classifying the process as stable or unstable. And to answer your second question on the scalability aspect. So we have deployed this model within our.

Server in the institute and I think 3 to 4 researchers are using and We have deployed on a machine with 64 gigs of ram and 4GB graphics with an I513 Gen. processor and 64 bit operating system. So till now we have not encountered any issues as multiple people are operating simultaneously. But if we deploy it on an industrial level scale, it needs one or two iterations to fix the multiple.

Of input data and we need to just calibrate it one or twice. It was the little work is needed when we deploy at an industrial scale and however the industries the latency should be low and as we provide it sufficient memory and the space to compute it, I think the latency issues can also be solved.

Thank you for elaborating, Professor. Your attention to ensuring uniformity in the input structure for model reliability is commendable, especially in achieving consistent results across processes. Similarly, your approach to scaling and accounting for potential issues during industrial deployment reflects a practical and thoughtful vision. One last question in this vein: If granted this role, how would you integrate your computational modeling expertise and AI-driven approaches to develop new research or curriculum that prepares students for the evolving demands in this field?

To answer your question, I have already made my own 2-3 elective courses that I would like to offer in VIT. One such course is Digital manufacturing where we take an example of a simple leap machine starting with the conventional machine, then how to convert it to a semi automatic, then how to convert it to a fully operational CNC machines and how to make it to a one step closer to a autonomic.

So these aspects I would like to offer to my students. And also I would like to establish an interdisciplinary research lab and work on the interdisciplinary projects that involve students and faculty who are more keen in exploring the aid-driven smart manufacturing, digital twins, process optimization, and which fosters the innovations at VIT. Also, I would like to incorporate some real-world case studies from my research on micro milling, charter detection, predictive maintenance.

To bridge the gap between academia and industry and this could be done and I will do it because I can leverage my experience that I have in the cloud based platforms and the enable learning to introduce a scalable real time monitoring and control systems as moreover on the research team. So and I would also like to mentor my students and fellow colleagues if needed in terms of publications patterns and conference presentations as I have.

3 patents out of which one patent is granted on the design, one patent is published and one patent has been filed. And overall, my goal would be to create a vibrant and an interactive learning and research environment that advances computational modeling and prepares students for future technological challenges. And moreover, I would like to focus more on the GPO accelerations also where we would be less dependent on the hardware and provide more.

After computational solutions.

Thank you for the comprehensive response, Professor. Your proposed initiatives, including introducing innovative courses like Digital Manufacturing, establishing an interdisciplinary research lab, and leveraging your patent-driven expertise, clearly reflect your commitment to advancing both academic knowledge and practical applications. Your plan to incorporate industry-relevant case studies and mentor students aligns well with the broader goal of preparing the next generation for real-world challenges. The emphasis on GPU acceleration and scalable computational solutions further showcases your forward-looking approach. This concludes my current line of questions on your professional and academic background. Let us know if there's any specific area you’d like to explore further right now.

Sorry to interrupt, we're down to the last 5 minutes.

So specific areas you mean to say in my research?

Not necessarily, Professor. I meant if there are any particular aspects—whether in your research focus, teaching philosophy, or vision for interdisciplinary collaboration—that you feel should be emphasized or further discussed, please let me know. I'm here to ensure we cover the key facets of your candidacy comprehensively.

Yeah, I'm interested in having a little bit more deeper discussion if you want on the how I would use computational methods in different manufacturing process. Let us can we can take a case study of additive manufacturing process or hybrid manufacturing process. So now currently my focus is on combining the additive and subtractive methods to optimize material properties and production speeds.

It's an hybrid active manufacturing process.

Your focus on hybrid manufacturing—combining additive and subtractive methods—sounds both innovative and timely. Let me ask you, then: How do you envision computational models playing a role in optimizing such hybrid manufacturing processes? Specifically, what unique challenges do you foresee in integrating these two approaches, and how would computational methods address them effectively?

See the problem with the hybrid manufacturing process is what kind of output that we want to get from the models. So whether do we need to analyze the subtractive part or the additive part. So here the challenge comes, what should be your input? What should be your output?

For a simple case study, let me say I am depositing on a simple mild steel alloy.

And I want to develop A5 M vertical one with multiple layers. So how I can use computational methods here? So the ideal plan is to deposit a single layer, then use conventional machining, make a rectangular wheel, then deposit the second layer, do a conventional machining again, then do the deposit and do the success shape operations. And where I can bring the computational methods here is in terms of prediction of the melt.

What is the melt pool width or the prediction of the beard geometries post the deposition prediction of residual stresses? Because these things helps me to characterize better on the product develop and this reduces the post processing characterization. And I have also in contact with a professor called Mr. Doctor Gundam ***** Kumar from IIT Madras. He is extensively working on this GPU solutions for microstructure predictions so I can collaborate.

With him to say that and now I'm working currently on this ML model prediction of melt pool behavior, residual stress and so on. Can we work together and combine for the microstructure so this entirely reduces the post processing aspect. So this is how I would like to work on.

Thank you, Professor. Your approach to integrating computational models for melt pool prediction, bead geometry, and residual stress analysis is not only methodical but also demonstrates a clear focus on reducing post-processing overhead, which is critical for the efficiency of hybrid manufacturing processes. Additionally, your mention of collaborating with Dr. Gundam Kumar to explore GPU-enabled microstructure predictions signals strong interdisciplinary potential and an ability to leverage expertise effectively. If I may ask one further follow-up, how do you plan to validate these predictive models to ensure their reliability and accuracy in such hybrid processes? What methods or benchmarks would you employ?

See validation always uh starts with the experimental methods and there is no rule that you need to validate. With experimental you can also validate the outcome which already published literatures. So I prefer to follow both the approaches. I want to do experiment, validate my model.

Again validate with published literature from the different materials. So this shows that my model is capable of understanding the process in a better when compared to the conventional approaches. And this also shows the models strength when it is an independent of the process.

And that's how I would like to show my repeatability and reliability by validating it with experiments with other published data, again with published data multi material aspects.

Your thorough approach to validation—combining experimental data with comparisons to published evidence across multiple materials—demonstrates a solid framework for establishing the robustness and reliability of your models. This layered methodology not only ensures comprehensive accuracy but also highlights the versatility and adaptability of your computational methods. Thank you for this in-depth discussion, Professor. Is there any other aspect of your work or vision for the role you'd like to discuss further?

Now in this current aspect, I would like to work morely on the hybrid and more. And I would slightly say that my methods and processes are not only the process oriented, I am more interested into development of products. So if my models, computational models are working fine, then I would like to connect with the commercial software industries and to say them that you launch these as a single aspect module for because we have.

Tested in multiple cases, so I'm more into product development rather than the process development.

So if you want to ask me questions on the product development, I would be happy to address it.

Hello.

Yes, Professor, I’m here. Would you like to discuss your approach to product development and how computational modeling contributes to it? If so, I can tailor my questions accordingly.

Sure, please.

To begin, could you elaborate on how you would transition computational models into fully developed commercial products? Specifically, what steps would you take to ensure these products are both functional and market-ready?

So again as I said you earlier, I would like to follow the two step validation approach, one with the experimental approach, one with the validation approach. So as I join as a faculty there I would be getting some doctoral students or masters students or an undergrad students. So I would like to ask some undergrad students to work on pre literature or published literature validation and I would like to ask master students to do work on experimental.

Aspects with some different combinations of material depending on the number of students I will get and if I have a doctoral student, I would like to ask him to do one or two experiments on materials. Then I would ask him to consolidate over a large variety of engineering material cases and develop a database out of it. So this is not a weekly or monthly process. It takes and one year to three years of work as we need to have a large amount of databases before going for.

Realization. So I would like to develop the database, rein trade the process again and again for multiple materials, retune it, make it adaptive for multiple materials and then go for commercialization.

Thank you, Professor. Your systematic approach—leveraging students at various academic levels to validate models, build an extensive material database, and refine iteratively—combines academic rigor with long-term commercial vision. This collaborative and tiered methodology is well-suited for developing robust, market-ready products. One question: When it comes to commercialization, what strategies would you employ to collaborate with industry stakeholders to either license or integrate these tools into existing manufacturing ecosystems?

So there are two approaches from my side. First, you can I can go for a patenting of it. And once the company says I'm interested in it, I can sell my patent to them and take some royalties on that. Or in the second aspect, I don't want to go for a patent. In that case, I would like to connect with the industry at my midway stage and show them my progress and tell them my future plans and I would ask them to collaborate with me and.

So once they collaborate, I would also get access to the industry part. Then both together we can take it further to the next step.

Both pathways you’ve outlined—either pursuing patents for licensing royalties or forming collaborative partnerships with industry during development—are judicious strategies. Engaging industry early not only enables practical feedback but positions your work for seamless adoption within manufacturing ecosystems. Thank you for sharing your insights, Professor. Is there any particular aspect of product development, commercialization, or research translation you would like to expand upon?

Time is up! Ending the interview now.