 Terry, welcome to Odea City, India. Pleasure to have you here. Now you're giving a talk. Can you tell us more about what you're talking about this morning? So we're going to be talking about, I am CEO of DeepKafa.ai. So we're going to be talking about the importance of research in deep learning and how do you take the fundamentals of research of deep learning into and applying it into the industries. So the talk is tomorrow, by the way. And I did give a workshop yesterday, a whole full day workshop on the same topic. And how did the audience receive that? Like, how did you find the audience? So I was actually really impressed. I mean, when we went into this workshop, you know, you tend to go with the script and you want to, you know, teach the students and you have a student community. And here we had people from various backgrounds, disciplines. There were, you know, professors and researchers and people with great academic background and, you know, people also from business domains. So in fact, I kind of really adapted my workshop immediately by what I asked these people what they wanted from the workshop into a couple of sections and was trying to satisfy the audience. So demanding but extremely well informed audience. I was blown away and, you know, we barely had lunch, but I mean, we were super excited. Well, of course, because deep learning research is such a hot area. Can you tell us a little bit about some of the areas you think are very exciting and deep learning at the moment? So right now I think you've seen, you know, computer vision and all this, you know, face recognition and a lot of things which, you know, requires computers and the machines to be able to see and make predictions and probably also inferences of what is happening are the areas we're working on already where we are also cooking, as we say, you know, in our labs and in our kitchen or algorithms and even a couple of really improvements in deep learning algorithms where we believe that we can improve these algorithms. So we work with OpenAI, we have, you know, we also have good relationships with Google and a bunch of other top researchers are joining our firm. So what we're trying to do is actually improve deep learning from a research perspective and try to create a robust kind of framework that is flexible, automated, and can, you know, adapt to your needs. So that's kind of a high level description. Tomorrow in my talk I will give kind of a deep dive description into what exactly this is and how do you really apply it and take it to the industry. We believe that taking research to the industry is the only way this industry is going to expand. We are concerned a little bit that research in itself becomes a great sort of a self-congratulatory kind of an exercise and taking it into the industry is where you really see the benefits and the fruits of, you know, the labor. That's very interesting because deep learning as I said very hard at the moment but a lot of great work been doing in computer vision, speech recognition, text translation. But do you see other areas of industry that can avail of deep learning models outside of those couple I mentioned? So obviously I think we could probably turn the question around and say can these technologies whether it's text, audio, video and obviously images and stills and pictures that you see on internet, can we apply these algorithms and technologies into different industry domains? So I think first of all I think the combination of these few disciplines. So you have computer vision based on audio, video text data and all these need to be combined and kind of make some meaningful conversation, meaningful understanding of a scene. For instance, a gentleman before me described about vehicles in front of a shopping mall. You could kind of make several correlations about how to take this into the industry and eventually put it into a different industry like agriculture, farming, marine engineering. We signed an MOU, I was mentoring a bunch of startups in London last week with the foundation. I don't know if you've heard of that. And so it's foundation Rebel Clinton where we're going to be meeting him next month as well in New York where you give a seed funding of a million dollars to a startup and they were from all industries, farming, wind energy and so we're kind of pulling those guys in and saying how can we expand research which is being today applied just into face recognition, you need to move it to different areas and do soil recognition, seaweed sort of farming and try to understand various scientific components of water, salinity and what else can we do. So I think you need to expand this into various areas. Yeah it's very insightful and so many possibilities there. So people looking to use deep learning in their work. What are some of the tools you recommend? You're working on a platform yourself, but what are some of the most popular tools out there at the moments that people are using? So I think to sum it up our conversation, we're bringing in a bunch of tools and actually also our own tools but you should look at TensorFlow and even PyTorps. TensorFlow is from Google PyTorps, Facebook's research is really focused on improving PyTorps. But also take a look at Chainer and a couple of other things. MatConnet is used as well in many areas, in industrial areas people don't talk about it. And just Python is a great language but people should learn also C++ again. It's a bit of a hurdle but the application of language that compiles at hardware level, at device level is something you need when you're driving an autonomous car or you're doing other applications. So it's important to understand the fundamental of using languages that really find their way into devices and material sciences of everything that we are seeing today. And then eventually using these libraries like TensorFlow, PyTorps, Chainer, MatConnet and all these other variations and get the best out of that. That's an excellent point because people tend to forget when they're using these tools they're so easy to use that deep learning is very expensive computationally. So Terry, how have you found ODSC India so far? Just simply amazing. I mean what Nareesh, Jain and his team have put up, I'm just blown away. And this is really no sweet talking or sugar coating because I'm very selective about speaking at conferences. I'm here, I'm meeting peers, researchers who are also presenting from biomedical engineering in other areas but these people are just too good. I mean, you know, I know they're throwing really massively dangerous curveballs at me but that's exactly what we need. We don't need self-congratulatory conferences. We need challenging statements and questions that drive us to improve our systems better. Well said, Terry. But I must be said we've got a very good turnout here and that's thanks to speakers like yourself, which are drawing in the audience. Thank you so much, Terry. Thank you so much, James.