Good day, Kushagra Tiwari. 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.

Hi, so my name is Kushagra and I'm currently working in ANSYS or currently acquired by Synopsis as Senior Engineer RND division my work focuses on.

Accelerating the finite element workflow for quick and fast data retrieval and currently we are working on a AI agent also known as MAPDL GPT agents for the finite element workflow. Prior to this, I completed my PhD from IIT Bombay and Monash University where my research was on data science coupled multi-scale mechanics for understanding for TVF and 3D printed products.

And I also have experience in academia for around 3 years where I have taught in state government engineering college.

In curriculum development, teaching courses and upgrading the lab facilities. And mainly my research focuses on integration of the physics-based model with the AI and current large language models so that we can have a better decision making.

Professor Tiwari, that focus on integrating physics-based models with AI—including the use of large language models—is a forward-thinking approach that holds significant promise for both research and teaching advancements. Could you elaborate on how you would integrate this focus into curriculum development? Specifically, how you would introduce these cutting-edge techniques in your courses or upgrade lab facilities to support such integration?

Yeah. So this area where, uh, the physics based model are integrated with AI is a very trending or very useful topic.

From my perspective, the area will be like to develop a AI agent where user queries something and gets his answers. So inside what will be happening is that we will be having many different models which will be used as a tools. For example, one tool will be to predict the uh, texture of the material. One tool will be related to the crystal plasticity analysis, One tool will be related to how the fatigue failure.

Seven tool may be related to the fatigue indicators parameters and one tool can be related to the fatigue life like how much material life is left at that current point. So when a user queries something. For example when a user queries, how much the material can can be deformed or how many cycles to failure is still left where the material fails. So the.

Yeah, Agent will call the necessity service and the necessity service which is in fact a code written using the physics driven Phenomena's will give the answer and the user will get it results. So this is the overall structure of the research. Now we can frame it into courses where one course can be related to the integration of AI with the LLMS and where we can use more domain specific knowledge and also in the from a lab perspectives.

Encourage students where once they develop the model, they start relating that to the AI and how that can be done where where they need to be like.

From development a lab we can.

Uh, very closely work with the AI agents which are based on different frameworks like land graphs or.

Uh, uh, uh, rag or those sort of things. So in this way, uh, student can like relate the fundamental physical concepts with the AI agent or work towards the, uh, like integrating the physics with the.

Artificial intelligence intelligence technology that is currently going on in the industry as well as in the academia.

That's an ambitious and well-structured approach, Professor Tiwari. Your vision of integrating physics-driven models with AI and involving students in this frontier of computational modelling holds tremendous potential. To clarify, how do you envision balancing theoretical instruction with practical, hands-on student engagement in such a setup? Specifically, how would you scaffold their learning to gradually build complexity—from learning fundamental physical and computational concepts to integrating them into functional AI-driven agents?

OK. So the very first thing that we need to be sure is we have a solid fundamentals in physics because that is the base of all the computation and all the results which we will be giving back to the user. So to develop a solid fundamentals, we need to be very good at the fundamental subjects of mechanical and computational engineering involving strength of materials and machine designs and then further on.

Build on that the students needs needs to be very good in the computational field where there are different algorithms like crystal plasticity. Once a student have the theoretical understanding of what the backbone or what the basics of the subject are then my next step will be use some hands on projects where a student can combine those with AI and how I will achieve this achieve that I will like encourage.

Which interdisciplinary approach or a collaboration between students of different branches? For example, a collaboration with students of mechanical engineering with a student of computer science engineering, a student for computer science can help mechanical student to understand how large language models behave and how MCP can be established so in this way. Mechanical engineering students will get a hands on experience.

AI and he will, or he or she.

I'm sorry he or she will be able to.

Correlate the domain knowledge with accelerating its physics physics in form model for a very quick result.

Your approach is sound, Professor Tiwari. By anchoring the curriculum in solid foundational physics and mechanical engineering principles, you create a strong base for students to build upon. Encouraging interdisciplinary collaborations, especially between mechanical and computer science students, reflects the real-world dynamics of such projects and can foster innovative problem-solving. Thank you for sharing this depth of thought. Let’s now discuss your specific expertise in computational modelling. How have you advanced computational techniques, for instance, in crystal plasticity or other domains, to produce innovative results?

OK, so the main area where I was working in my PhD were a multi scale computational modeling and the innovative thing which I've done is related the fatigue failure model at two different length scale. So the approach or the research that I carried out was published in one of the highly reputated journal that is International Journal of Plasticity. It is a very highly reputed journal and.

My work was published in that and what I did is I related a.

Micro skill model which was based on crystal plasticity to identify the fatigue failure hotspots which are also known as FIPS and from that I related those with the fatigue life of the component. That is a macro scale model and these two model are at two different length scale. Micro-scale model is based on the slip systems and how the deformation is occurring at a micro scale where there are micro.

Involved in crystal plasticity takes care of all those things whereas.

At the macro scale, we are interested in the crack growth rate and I use equations like Hartman equation to model the crack growth rate. Now what I did, I related the crack origination.

No, using the crystal plasticity model and those the parameters of those model, I related it with the macro scale model of the grad growth rate. So in this way I developed a unified approach starting from the grain scale or the meso scale to the macro scale, a sort of a structure property relationship in a material materials engineering perspective.

And I also used data science concepts for this or.

Uh, like it, uh, the topic I will reconfirm or reiterate was data science coupled multiscale mechanics for understanding fatigue behavior. And this process or this investigation was done in 3D printed or additive manufactured product titanium alloy. So this was a novel work and this was the research which I did was the first research that correlated, uh, fatigue failure models at different.

Skilled and produced, uh produced results which were validated using the experimental data.

Your research on multiscale computational modelling of fatigue failure in 3D-printed titanium alloys demonstrates a pioneering approach to bridging micro-and macro-scale phenomena. The integration of crystal plasticity at the micro-scale with fatigue life modeling at the macro-scale—validated experimentally—is a significant contribution to materials engineering. Publishing in a high-impact journal like the International Journal of Plasticity further reinforces the credibility and importance of your work. How has this research influenced your teaching or mentoring approach? Specifically, how do you translate these innovative techniques and findings to inspire students in theory and lab courses?

Uh, this research has helped me immensely.

Now, since I know I've worked at a particular problem in a very deep way, I know what are the problems or what are the practical challenges that exist.

Earlier I was just more focused on a theoretical background, but after this research now I like relate the theory with the practical problems that a student may face when he or she like starts doing a project.

So this research has helped me to give students a very deep insight, give them a research perspective and also give them a industry stranded or the practical challenges that they may face apart from the theoretical knowledge.

Your ability to tie research with both practical challenges and theoretical foundations provides students with a well-rounded and enriched learning experience. Equipping them to address real-world challenges while understanding the depth of the theory behind them is instrumental in shaping proficient and innovative engineers. Let’s turn to your interdisciplinary approach. Could you discuss how you would guide student projects and research that incorporate computational modelling, AI/ML applications, and materials science, aligning them with industry-relevant outcomes?

Yeah, 1 short example is in the research which I did myself where I related the physics model at different length scale and the process how I related the micro and the macro scale was using a data science concept which is specifically Bayesian modeling. So I use patient modeling or data science concept to relate these two scales so.

I always teach students that we cannot do a research only keeping in mind a single engineering department. It has to be interdisciplinary and the ways which we can use.

Is to collaborate with the current trending topics that is a I where we can use the large language models which are trained on a generic data to be get trained around a core mechanical engineering data and fine tune them so that we can get better engineering decision results. So the idea is not just to develop. A simple pipeline for our AI agent, but develop it which has a very.

Good domain knowledge, uh, from a mechanical or a materials engineering point of view so that we can integrate materials engineering with the AI concepts.

Yeah.

Understood, Professor Tiwari. Your interdisciplinary methodology and focus on integrating domain-specific frameworks with AI demonstrate a clear vision toward advancing both research and teaching. Let’s now focus on your ability to teach advanced courses. How would you structure a course, for instance, on integrating AI with computational modelling for materials science, considering both foundational and cutting-edge elements?

So the course will be, uh, divided into way many categories. The first part will be to build the foundation of both the computer science related or the artificial intelligence or the large language model related basic mathematics involved in that because it also has various.

And the second part will be the building, the foundation of the physics involved phenomena which is core specific to the materials engineering or the.

Components like vector algebra or calculus. So building the foundation in that.

Mechanical engineering. The third part will be re correlating between these tools. How we can relate these two things so that, uh, there are better engineering decisions that we can make. The 4th part will be to.

Uh, make this.

Application oriented related and the 5th will be the scope of further or the future development. So using these five categories, I will cover the foundations, the basics, the challenges involved, how the correlation can exist, what are the industry and the practical issues that we can face and lastly, what are the.

Prototype a more application orientation.

Possibilities or in which direction a particular student can go on if he or she wants a more deeper research or if he wants to deep dive into this subjects. So these five categories will cover from zero to 100% efficiency of the subject.

That’s a well-structured approach, Professor Tiwari. Starting with mathematical and physical foundations, moving toward integration and application-focused prototyping, and concluding with future research directions ensures a comprehensive learning pathway. It empowers students to not only understand the subject deeply but also navigate its application and innovation scope effectively. Now, shifting slightly, how do you plan to evaluate the effectiveness of such an interdisciplinary course and ensure that students are achieving both theoretical mastery and practical proficiency?

So I believe in peer learning and continuous assessment and.

Daily quizzes or monthly.

UH talks which are organized by the state students for the students where a student can talk on a particular topic every week. So I do not believe in a assessment one time a year. The assessment should be continuous so that I can track how each of these student is progressing in their subjects. So I'm more focused on a continuous assessment and peer learning sort of assessment and.

Meaning practical knowledge by giving students a some hands on project instead of just writing theoretical answers on a notebook. So these things first the practical knowledge and hand association, second is the peer learning and third one is the continuous assessment that are the primary focus through which I will try to make students learning very easy and very.

Beneficial for them.

Your focus on continuous assessment, peer learning, and practical engagement reflects a modern, student-centric pedagogical approach. It not only ensures consistent progress tracking but also fosters collaboration and application-based understanding, which are critical in interdisciplinary fields like computational modelling and AI integration. Let us now discuss your ability to contribute to research development. How do you plan to engage in publishing high-quality papers while also maintaining a collaborative and innovative research environment within the institution?

I plan to extend my current work in PhD. My work was more of a like physics informed work and my current company it is more of a AI related or AI agent related work. So I intend to work in a direction where I can combine or integrate these two phenomena to publish them in a very repetitive journal, uh, like International Journal of Plasticity, Computational material Science and journals with very high.

Big goal cannot be achieved alone. We need a strong team or very focused team. So I am like, I firmly believe in collaboration. And from my research, I, I mean, I will be interested in collaborating with people from mechanical engineering, from materials engineering, from computer science engineering.

Factor and for this I will be collaborating because I believe uh.

To work in a collaborative manner to achieve a very desirable and results which are helpful for the Institute.

Your vision to integrate physics-informed modelling with AI-based approaches demonstrates your commitment to advancing cutting-edge research that holds both academic and industrial value. Your dedication to fostering interdisciplinary collaboration across engineering domains aligns well with the goals of nurturing a strong research environment within an institution. Thank you for sharing your insights so far. This brings us closer to wrapping up my evaluation. Is there anything specific you'd like to address or highlight further before we conclude?

Uh, yes. Like, umm, if I want uh, to present my work at a different conferences or if I want to collaborate with people other than.

The institute Am I allowed to do that because I am strongly believer in collaboration and getting a very high results from my research. So is is it possible?

While I am not able to speak on specific institutional policies, universities and research institutions typically encourage faculty to present their work at conferences and engage in collaborations beyond the institution, especially when such activities enhance the visibility of the research and the institution's reputation. Collaborative partnerships, when aligned with institutional guidelines, can be incredibly impactful in driving innovation. If you were to join, it would be advisable to discuss these matters with the administration or department leadership to align your goals with institutional goals. This ensures mutual support and aligns resources for your collaborations effectively. Any other thoughts or questions you'd like to raise before we complete our discussion?

No, no. Thank you for answering that. Uh, no other questions from my side at this point of time. Thank you.

Thank you, Professor Tiwari, for sharing your insights and expertise throughout this conversation. It has been a pleasure discussing your research, teaching methodologies, and vision for interdisciplinary collaboration. Your depth of thought and innovative approach are commendable. If you've confirmed that you're ready to conclude the session, kindly let me know explicitly so I can finalize and close things formally.

We can conclude. Thank you.

Sure, sure. Yes. Thank you.

Just to confirm, are you explicitly stating that you’re ready to end the interview? If so, I’ll proceed to close the session formally.