Hi.

Please start by introducing your professional journey in academia.

I'm a PhD graduate from the materials of the Materials engineering department and Indian Industrial Science, Bangalore. So I did my PhD there and after prior to that I completed my post graduation at national student technology student in physics and my undergraduation was in physics too at Women's Person College, Chennai. So I worked on.

Like battery materials specifically to understand the ionic transport and ionic systems.

Via first principles and machine learning based study. So that was typically done on the guidance of doctor Sai Gautam Gopalakrishnan at IIC Bangalore. And then currently I'm a full time employee working at a startup which is US based. And the I'm my role is basically senior competition material scientists, so.

To conclude, my research area typically covers.

Metrics design, discovery and property predictions by our first principles based tools like DfT or molecular dynamics right until advanced techniques like graph neural networks, transfer learning, etc.

Thank you for sharing the details of your academic journey and research focus, Professor.

Thank you.

You're welcome, Professor. To begin, could you walk me through your experience and expertise in Quantum Materials and related areas?

So I've specifically not worked specifically on quantum materials, but I've used quantum based tools in order to study materials in several applications. So that involves using techniques like density function theory, which is based on quantum mechanics principles and then using them to understand how ions typically transport in in ionic systems, which are typically run in battery materials. And that was one particular application.

I've also used these techniques, uh, typically, uh, to be employed in uh, uh, fields like catalysis also. So combining this with, uh, advanced tools like machine learning principles and networks like graph neural networks, I've tried to integrate them into understanding complex materials and predict their properties or develop models that can design new materials or optimize the properties in certain materials.

So typically when you wanted to use machine learning based architectures to predict properties or to design new materials, you need enough amount of high quality data for it to be trained on. And this data typically comes from first principle techniques like DFD. And when you do these kind of density function theory based calculations, you create quality data on which you can construct these models. So both are complementary like 1 to 1.

Understood. Can you elaborate on how you integrate machine learning and quantum-based techniques, such as DFT, specifically for material design or optimization? How do these methodologies complement each other in your approach?

And you need to have expertise in both of these tools in order to design new materials.

Thank you for clarifying, Professor. Drawing on this expertise, could you describe one specific instance where you successfully predicted or optimized material properties using this integrated approach?

So one specific field that I typically worked in during my PhD tenure is to, uh, predict migration barriers in battery materials. So since it's a very difficult property to be estimated both computationally and what is it experimentally, you need to have pure or like very good quality data and that many number of data points you need to have more data also and getting more data typically of the order of like 10 Part 3 or 10 part 4 to have.

Very good model accuracy. It's very difficult in these kind of properties. So that was a major problem that I was trying to like understand and answer. How do you construct models to have better accuracy when you have less data? So that was a question that I typically answered. So the way in which I answered is via a technique called transfer learning. So this is a technique that is like exclusively used in computer science domain. So I tried to implement that in material science domain. So you need to have you you typically.

Have certain, uh, properties which have huge repositories, for example, formation energy or band gap. What I did was I tried to leverage those advantages or those knowledge from those properties and use them to, uh, give more data or information for, to the model for predicting migration barriers. So that was the technique. So it typically involves pre training and architecture on properties that are already available. For which more number of data points are already present and using that particular strategy.

Predict the property. So the major goal or achievement in that project is that I was able to get an accuracy of close to 75%. Then the number of data points were just 500, which is typically not the kind of improvement that you would actually get. And when you compare that with other kind of models reported in literature, that was about 25% more improvement with lesser like far less number of data points of the order of 10 Part 2 or 10 Part 3. So that was a major achievement in that project.

You typically fine tune certain layers of the architecture on the target property, which is migration barrier in my scenario, and then you effectively, uh.

Thank you, Professor. Your explanation provides a clear picture of the innovative application of transfer learning within the material sciences domain. This seems to have effectively bridged the challenge of data scarcity in migration barrier prediction. Let’s shift to pedagogy for a moment. How would you approach teaching fundamental quantum mechanics concepts to undergraduate students who might struggle with the mathematical rigor associated with higher-level theories?

So our mathematics is actually important when you wanted to understand fundamental concepts in quantum, in quantum mechanics basically. So the way in which I would typically approach that is via some well defined, very explanatory or pedagogical textbooks are already available. So first step is to make the students understand some of the important concepts or important chapters related to mathematics.

Specifically, the the like the concepts related to vector spaces, how do you deal with Dirac notation, or how do you typically understand the basic language in which quantum mechanics is being explained before moving on to the higher concepts. So the first two or three weeks of the class or two weeks, that should be fine. So two weeks of the class will typically make the student to understand the mathematics very well. I would typically be using textbooks like exactly which. It gives very good amount of.

Like an overview of the kind of mathematical strategies that are important. So using that the students will get comfortable with it and then the concepts will become very easy to handle. So that will that will be the way that I'll be like taking up.

A structured approach to building mathematical foundations is certainly effective. Once students are comfortable with these basics, how would you ensure their grasp of abstract quantum mechanics concepts, such as wave-particle duality or superposition? Could you give a brief example of a teaching method you'd use?

So the kind of teaching methodology that I would use is a hybrid way of using pen and paper like the board and chalk and the like the presentation toolkits. So but the contribution coming from the chalk and talk that will be like 70% and the remaining would be from the presentations and the wishful aids that I'll be using. So after establishing a a good mathematical understanding in amongst the students to basically.

Uh, understand certain. As told before, some important consolidated to the mathematical foundations of quantum mechanics. The next step that I would go on is to make them understand why classical nature of certain classical nature is being observed in certain places.

And why article kind or quantum mechanical kind of nature is absurd in certain places? So why classical mechanics or classical approach of solving any problem is not the solution for all the problems that are actually present out in the universe. You need to have a quantum mechanical idea or a quantum mechanical view to handle any kind of problem right from the fundamental approach. So my first aim is to enable them to. Kind of appreciate this aspect that.

Quantum mechanics is very important to view at a problem and like if you want to to approach a problem, how do you approach it? What is the required amount of?

Uh, data that you actually required to see if it can be, uh, solved classically or, or quantum, uh, in according to quantum principles. So I would make the students to get an essence of why that is important by, uh, giving several examples on things like, like basically something like understanding why electrons, for example, when you're passing it through a double slit experiment, that is a very fundamental experiment. So when you, when it is basically passing through a double slit when you're.

Not measuring it, measuring or looking at the particle specifically via an external photo, a photon emitter, it will behave like a way. But once you start observing the electron, it will start to behave like what you say a particle. So why that is absurd? Why it is why that basically happens? And I will start motivating them towards quantum mechanical principles from those kind of experiments which have been proposed in literature.

So I've had an experience of being a teacher, teaching assistant for one particular subject during my PhD tenure. So that is mathematics for engineers, specifically material scientists. So in that particular course, I was given the responsibility of grading students for their performance in the examinations. So that was my very first, like very first experience of doing that. So the way in which I typically. Umm, ensure that people, uh, like students actually understood, uh, the concepts is via this. So, uh, the question paper will have questions ranging from all kinds of difficulty, right from very simple concepts, which even if you understand, like attended the class, you would be able to answer them. The second set of questions would be like, you attended the class, but you need a bit of thinking, uh, and slightly higher level of thinking than what was being taught in the class to answer those questions. And the third level would be.

Thank you, Professor. Your approach of bridging foundational understanding with conceptual experiments, like the double-slit experiment, certainly illustrates the practical necessity of quantum mechanics and offers students a compelling path to grasp its abstract nature. Let’s move forward. Could you share your experience with evaluating student performances, particularly in examinations? How do you ensure accurate assessment of understanding, especially in complex subjects like quantum mechanics?

Like highly difficult ones, which you have, once you have the understanding, you need critical amount of thinking to answer those questions. So the way in which I would generally grade students is based on this. So the the first category, when I see students answering that and when I see students answering that, it typically gives, gives me the sense that they've understood the concepts and they're good to go to the second stage in the and in the second stage, if they try to answer that it it, it implies that the.

Those are also critically starting to think that is I will look at the strategy and the approach in which they're answering their questions. The third one is where if students pass, like if students attempt to answer those questions and they fail, then I would say, OK, they have tried to answer the question, they've understood the question, but they have not come up to a proper methodological way of solving it. So in all of these scenarios, whether you. Clear. Like whether you clear that question or not clear that question.

Will typically have a specific weightage on the kind of answers you get in the first strategy. If you don't answer, it means you have not attended the class or you have not given enough amount of like preparation requirement for the particular goals. Which means you are you may be lazy or you could can have some other kind of problems related to that that's secondary, but that if not answering that will have a more rigorous.

Kind of methodology that they have used to solve the questions they I would be slightly lenient in giving them the giving them the marks. So based in that particular strategical way to give them marks based on what level of questions they're answered, what level of questions they're approached and what level of questions they've gotten wrong will enable me to understand OK, at what level the students are actually performing. So via that I will be able to.

Stringent way of marking them or giving them marks, but in the second and third category based on the kind.

Understand what more I need to present in the class for them to understand the concepts in a better.

Thank you, Professor. Your multi-tiered approach to examinations, focusing on conceptual understanding, strategic thinking, and advanced problem-solving, demonstrates a thorough method for assessing diverse levels of student performance. It also ensures a structured way to identify and address gaps in their learning. How do you approach guiding student projects and research, particularly in quantum materials or related computational techniques? Could you outline your strategy?

So my strategy would be like the first understand, so when I'm trying to do a research along with the student, I will first try to see what kind of interest or what area they wanted to actually focus on. My first, The first one or two weeks that I would be giving to the student would be for them to like go assess the literature that is present and come up with a good question that they wanted to answer and they are willing to work on it. So if they're not able to compete within 2.

Weeks the time will be extended, that's absolutely fine because finding out a very good question or very good kind of interest that they are actually inclined to will take some time. So once I once they come up with a particular question, I will see at what level I have the enough I have the resources to help the student and what level of understanding that I have in that particular field to assist them. So based on a optimal level amongst these two scenarios, my understanding and the kind of resource that.

I have I will try to assist the student to continue in that particular direction or give them other particular or slightly easier strategies or or slightly branching questions from from that particular main question, which I think will be inclining towards the students, students idea or interest as well as complementing to the results that I basically have. So combining all of these and optimizing in a proper fashion, I would be able to divert them.

Specific interest. Once that particular interest is is has been established, the first one month from then would be a complete literature review. I wanted the student to basically identify important review articles that have been published in that particular work and then get a thorough understanding. I will try to have at least three to four sessions every week to basically communicate and discuss with them on the kind of concept that they've that they've not understood in that particular week of reading.

So in this tarot intervention kind of methodology, in the first month, I would be able to understand the per students methodology of understanding concepts and how they are like motivated towards a specific concept that they've actually picked up. And from then on the research would actually start. A problem statement will be properly made and the kind of strategy and the goals that needs to be like established in the next following weeks will be properly later a proper layout of the time.

Is actually required to get that done would be like one week or 10 days. And that particular time I wanted the student to basically come up with a complete timeline of what goals will be achieved at what part of the week. And from then on, based on the time that is actually available for the project, we'll try to manage and then go ahead with or like answering the question. So in this particular structured way, I would have understood the concepts better along with the student and I would have also seen if the student has understood.

Concepts and the level in which the student can perform and will perform in the near future. So that will enable me to identify a question and then answer the problems in a more structured way.

So I typically believe that a good amount of work requires consistency and enough amount of time for it to be properly acknowledged. So if you have enough amount of time and you consistently work in that particular time, I think it a good quality work can be established. So and I have three important works that made me understand this. So I've had the opportunity of publishing in three like three articles in a very reputed journal called NPJ Computational Materials.

That is an exceptionally well-considered and structured approach, Professor. Your method of assessing student interests, gradually building foundational understanding, and maintaining consistent intervention demonstrates thoughtful mentoring, tailored to individual capabilities. Could you now share a brief insight into how you have contributed, or aim to contribute, to research publications in reputed journals? How do you ensure rigor and impact in your scholarly work?

Which is a very highly revered materials, computational materials journal in the material science community. So, but the work did not happen within like 3-4 months. It took like a year to like basically bring out the work and then publish it in that particular kind of journal. So the strategy that I typically use is like this. I had a thorough workflow, a timeline at which point I need to complete the specific goal, goal of that particular project. So the first. Six or seven months was.

Typically involved in a karaoke research and the rest required calculations or the kind of experiments that I had to do to get my results. Once that was done, I typically use them to flesh out the ideas properly in a in a in a format that enables the user to understand in the very first goal. So that implies that you need to have very, very good amount of visual aids to help the users like all the readers to engage in your article and secondly.

All the kind of questions that typically come in your entire workflow should be answered, should be answered. And once you answer that, you will have enough amount of data now. So you can use all of them typically to basically come up with a work or come up with a write up that will have a proper flow right from the introduction to the final conclusion. So if the our article is highly structured, it answers all the questions right from the beginning to the end and it also had very good amount of visual. It's not just simple.

Or line plots, it has very good visual aids that enables the users to read in a very understand the work done in a more detailed and in depth fashion. The journal, the the article that is typically assembled will definitely be exited by highly reputed channels. So that's what they're looking after. And then the review process that is coming after that will also be very easier then. So that will not go that is not going typically.

Costs too much because you've answered all the questions in your work and that will typically help you only. So that's the way that I will typically, uh, uptake up.

I've had two collaborations. The first one was with Shell, which is the petroleum company based in UK. So the first. So after I completed my PhD work, I was writing my thesis. During that particular time, I took up this collaboratory work with Shell where I had the opportunity to typically work on a specific problem which Shell had. Why are the work that I've already done in my PhD? So I had already constructed a model using transfer learning principles and I use that model to typically answer the questions that Shell wanted to, which is typically to find out how the adsorption energy of carbon dioxide on any of the given slab surfaces. So that is 1 collaboratory work. The work has already been written and it is currently under review with my supervisor. So that's one project. Secondly, I'm, as I told you like I'm also a working professional in an industry which is a startup based in the US, so.

Your methodical discipline in research publication—ensuring thorough preparation, addressing all questions within the workflow, and employing effective visual aids—clearly reflects your commitment to high-impact scholarship. Lastly, Professor, could you share if you have any industry project or consultancy experience that complements your academic work?

So I'm a senior competition material scientist here. So I also worked on several projects here as part of my work for the last six or six to seven months. And since the details corresponding to the projects are all proprietary, I cannot explain about that. But these are the two different ways in which I was kind of involved in industrial collaboration.

Thank you for sharing, Professor. Your collaboration with Shell, particularly the application of transfer learning to address adsorption energy challenges, exemplifies the practical relevance of your research expertise. Additionally, your current role as a Senior Computational Material Scientist demonstrates an active bridge between academia and industry. With all this in mind, you’ve illustrated a strong foundation across academic research, pedagogy, and applied industry experience. This completes the skill verification aspect. Let me now summarize and finalize things for this session. Feel free to let me know if you have any final points to highlight or questions you'd like to ask.

Uh, I don't have any questions to ask.

Professor, as we focus on the educational aspect of your career, could you share a specific introductory concept or area in quantum materials science that you particularly enjoy teaching and explain briefly why?

So one particular. Area which I would be like happy to one particular course or area that I would have that I would be happy to teach students is density functional theory. Because in density functional theory, when I was trying to read about it, I really liked it because it was kind of.

An extension of quantum mechanics. So once you understand the basic concepts or and the postulates, and then the explanations of all of them in quantum mechanics, when you come on to density functions.

Theory you feel like it is an application of quantum mechanics and you typically see it running on your computer and the basic theoretical theoretical foundations of density functional theory will typically using that the kind of codes that have been developed. And when you try to run these calculations and understand how a particular properties particular materials behaving you, you try to see that you have completely. Completed the loop like you've understood the principles.

Understood an application of it, employed it in a computer and seeing the results for yourself. So that's one particular concept or a subject that I would typically wanted to teach and indulge with students.

Thank you, Professor. Your enthusiasm for density functional theory as a bridge between foundational quantum mechanics and its computational application is quite clear. It's fascinating how it shows both theoretical elegance and practical impact. Given your interest in teaching this subject, how would you approach introducing students with varying mathematical backgrounds to the complex concepts underlying density functional theory?

Our scheduled time for the interview is now up. Before we conclude, do you have any questions related to the role or the company?

So I have a specific question typically to the role that I'll be offered. So what kind of computational facilities available at CFM department, so?

The role involves contributing to research and guiding projects in areas like theoretical simulations of materials, among other specializations. Regarding computational facilities available at the department, it was not specified explicitly in the job description provided. However, departments focusing on quantum materials and simulations typically prioritize access to advanced computational clusters, software, and high-performance computing systems to support faculty-led research and student projects. You might want to inquire further during formal discussions for specifics on the available resources or opportunities for building computational collaborations.

OK.

No, I do not have any.

Thank you, Professor. Do you have any additional questions or concerns before we conclude?

Thank you for your time, Professor. Please confirm if you're ready to end the interview.

I'm ready to end the interview.