How tech professionals build aiml expertise
He asked and has a experience of almost three decades 10 in edtech, and almost 20 in FinTech and has been you know, driving business and marketing for talentsprint. Here to go to Paris. Good evening, everyone. Nice to meet all of you here. servi joining. Yeah,
we should make it a feminist when she comes in.
Hi, nice to see all of you here and good to have almost 100 people are already on the call. It's very interesting times that we are going to, I would say interesting, in other word challenging or difficult at this point of time, things are very different than what we are all used to. And I think it's important as professionals, all of us, get ready for what is happening, what is expected to happen ahead of us and be ready for it. And I'm happy to connect with all of you through this session. What I will As I know, with the webinars that a lot of video calls that all of us are doing in work, so there are some people who will come in too late, I expect 150 people to be in the session. I will start with the five minute intro about talentsprint. So that when we go dive, deep dive into the actual content of the webinar, people are already in. I am here trying to first
get my presentation
shared, just hold on for a sec.
I hope all of you can see my presentation, you can just confirm on the chat in terms of format, what we will do is I will take you through a few slides and then I have some questions we can discuss. And I also have analyst Surabhi, who has joined us who is who was deeply into this program, she will also help all of us understand things better. So Just about the structure, I'm sure you would have seen the main topic, the whole tech professionals with expertise. This is not something that I'm going to experience nor expect to share with you. It's based on the journey of around 2300 participants who have gone through the program. My objective is to bring those things together to you take your questions and inputs on taking over, and Surabhi will who has attended this program and wanted to further deep dive into msps and now is one of the mentors for the program. She will also be joining me and chatting how to use our entire experience of participants in this program, about talentsprint very, very short four minute update. talentsprint is just about 10 years old at this point of time. Our focus is completely on the deep tech areas. We do deep tech programs for both working professionals and graduate students. Of course getting us back the company started with graduate programs. But over the last three, three and a half years we have been focusing on working professionals. And our programs are all on technologies which are deep tech and evolving. Ai machine learning data science is one category. FinTech blockchain is another category. cybersecurity is another category. We have digital health, which is a recent program that we launched. And we work extensively in areas of robotic process and digital process automation. We work with organizations, which you would see in the next level. Our programs are largely in partnership with top tier academic brands are working professionals, and top Branson technology companies for the graduate students. So our program on AI machine learning with total it has robot, which is number one machine learning lab and number one computer science research organization in the country. The FinTech blockchain FinTech program is with I am Calcutta. cybersecurity with it kung fu digital health is the number one institution in the country which has Indian Institute of Science which we announced a month back during the lockdown RPA program is with RP a DPA program is with pega automation anywhere blue prism. And as a company we also run probably the world's largest diversity program for women tech professionals along with Google and we are at this point of time training 250 people and objectives to train 600 people from across the length and breadth of India and make them globally competitive for top tech kind of rules in the company. The company is largely led by a leadership team which is from the industry just I will not spend too much time on this. As he said, I have been from the financial services industry for long trying to lose IIT Madras graduate and our university of michigan PG started his career with Watson labs in IBM and is one of the cofounders and the CEO of the company, are shokan is considered probably the top Python guru in the country along with his experience in various areas. He does classes along with IIT Mumbai and others, and GTS, our CTO, Dave, Johnny x Microsoft, head of partnerships and Gurpreet, from Honeywell and PwC XLR glad, she is a HR person and he takes care of talentsprint Hr needs. We have good investors. I broadly, I mean very, very short, this thing into a game about these institutions and programs. So these are the top institutions we work with. And today our focus is on triple it hydrobath program on AI and machine learning. We will cover that this is the first program we launched a very, very interesting program. And these are the program for college students. That will do Okay, so that's the short update. Now we have 120 plus people on the call, I think, time for getting into deep diving into the topic of today's session.
Okay, so what before I get into this, I was trying to launch a short poll since underdog people are there. We believe that the requirement of a working professional for building expertise in new areas like deep tech areas, again, machine learning is very different from an academic programs that people do. So I have a very short 454 question, which will take less than 30 seconds for you to answer. I'll be grateful. It'll be coming up on your screen, and I will be grateful if you could answer it. I can take that answers and start the session so that I can put the content in context with this, I have launched a poll. It will be on your screens now.
There are a few questions as somebody who has
raised their hands, we will definitely take this questions. I'll be happy if you can respond to the poll that has been started and waiting for answers. The first person has just pulled up. Okay, now it's coming thick and fast. Okay.
30% other people have already done it.
That's very fast and touching 50 now 50% of the people. Mr. Narendra, I think you're not able to see are you connecting on the mobile phone iPhone or something?
If that's the case,
oh, it should be. I mean, I've been able to see it, maybe just refresh and you should be able to see the poll. We have around 80 70% of people who have already submitted, there are some who are unable to see it. Just scroll down and see I mean probably is not visible on the screen absence the number of questions or more the surgery and our current current has been able to build on mobile. Yes. I have seen that there's a little bit of issue in some of the iPhones But otherwise, things have been okay, I'll wait for I think more than 90 people have given I'll just wait for another 30 seconds to ensure that we have crossed the number 109 Stop it then. There are more than 120 people in the session. At least 100 polls is also be useful.
Thank you fine now. That
and just four out of hundred
and hoping to get three more We are at 9798. Now
Okay, you want to go
here we go.
Okay, I've got hundred. I'm now ending the poll. Exactly. We have use taken almost four minutes for the small. Very useful. Thanks. Thanks a lot for your contribution. I want to share those results but I just wanting to read it through once. So that
Good, pretty interesting. I'm sharing the results on your screen.
Plus, you're able to see it
Are you seeing the results on the screen? Yes. Okay.
The poll has been stopped. I'm sharing the results on the screen. Yep. Okay, so as you will see here, I think the first question, your program for professional should be very different from an academic program for students 79% or 80%. fully agree 17% partially agree and 2% disagree, right. Only 2% of the people believe that it would be similar to an academic program. 98% of the people believe that it should be different. Some of them fully agree with at some party, I believe. This is a very interesting insight, right. And this is something we as a company, experienced when we went online. So when we went to executive programs, right, and when we went to executive programs, we really found that there was a need to be very different and That's clearly visible from the first cohort of this program itself, we have been able to convert the new structure and we have been constantly improving on that and triple it has been doing very, very interesting changes to the program to make it exciting and interesting for working professionals. So that's very conceptual. Now let us look at the second one. The ideal approach for professionals would be I'm sorry, I think I'm to read it from a past screen. Let me just look at the screen, okay. Start with theory, concepts, algorithms, and then learn how to use them in the real world. Critically personal people believe that that's the approach while 76% of the people believe that ideally, we should start with real world situation, get to know the algorithm which can help them solve the same and understand the concept and theory behind it after that so that they can use it in the real world. Absolutely a very, very clear mandate towards the second option, right? Now let's look at the third question. professionals at different hierarchical levels and experience level will have different objectives in a lab session. Right. 72% fully agree 26% partly agree and 1% disagree. Okay. Now learning from peers is very important for a professional education and peer group will influence my learning 86% agree with it. 14% partly agree with it and nobody disagrees with it. Very, very interesting. I think it's I mean, this is something we have been experiencing for the 2300 people we have been going through. And I think the what you are showing here is similar to what we have experienced now and stopping the sharing of results. Now, there are a few questions. There are some Show of hands before I deep dive into this. I'm going to allow syniverse to go ahead, Seamus, you can unmute yourself and ask your question, please
Can you give us
your raised your hands? Yes yes yes I have reached please go ahead with your question triggers
right I'm just to be honest, I'm working on the AI. Yeah. And joint joint an AI company only like
on Lego the team is also to be honest TVs also not a technic, theoretically they are very well. Theoretically they are very well, programmatically and they include the MC part and all is a very difficult for me to lead the team basically. I'm just trying to understand from them, right, I know what they have learned from the YouTube or somewhere else. I get to record some different porters, right. End of the day real implementation part they can't able to do it. All 40 Absolutely, yes, that's that's a big challenge that being faced across all technologies, including AI, and ml much more, because you don't have a defined input process output right there. That's on its own. So that's that's the real thing. And the the persons who are able to manage projects, well, are those who understand how algorithms really work and which algorithms to use where the person who's really good at coding may not know this, a person who knows this answer to this question may not do good in coding, right. That's the real challenge and need for multiple levels in an organization comes a new we'll we'll cover this Yes, there's a very important questions and some of the slides that we will present going forward and also the results that we have seen in the poll that we just launched, will come in handy in that universe yet you're you're right in the know overall. Like and I understood that they So you are on mute certainly more about all basically the logical thinking they can't they're unable to do it. Basically, they are not thinking logically, they're coming back with a solutions with a different thought process which is not fit to the system. Basically, I completely agree with you the differences at this point of time and talking about asset not day. So how each one of us can build it in such a way that we are in, use them to come up in stage and go ahead. Hold that answers sreenivas I have one more thing he was wanting to stop today seems to be the day of cinemas. Let me allow you to unmute himself. Go ahead. Hello.
Please go ahead. Yeah, yes, yes, yes, yeah. Good evening, Mr. Schrader.
Nice to have you webinars. Thank you so much. Yeah. Yeah. So I am also working in the tech company like double IoT. The company's doing in double IoT. So my question is, until now, I have trained the models on toy data sets, right? So when we came into the company scenarios, the data is not relevant. That's true. Yeah, what we learn and even after,
so lots of effort of ETL.
And also we can't able to achieve the metrics, what we needed. And like, our reporting managers are having goals that we need. We need to deploy this model we need to deploy this model, but like, they don't want to do research. They just want to deploy their models. So and also like how this Like workflows, like nowadays, you heard about that Apache airflow, right?
Yes, yes, yes. Yeah.
That that do the automation in the automatically trains the model like cron jobs like cron jobs, right? So how to use that in the
real time scenario.
So that if I have to start answering it, number one, I probably not be not fully, technically ready for something like this. But having said that, the whole session will give you go in answering this specific question. It's true, there is always a challenge when you transfer an algorithm which works on a toy data to real data. And that's where the building of expertise at multiple levels of an organization is extremely important, both from the data cleansing and making it really side to how the company is going to procure data or not recollected, maintain and update data from the various touch points and how it can be used effectively. If you really look at it. Advanced organizations using AI have been successful in doing both sides of it, not just the algorithm development and implementation but data aggregation, data maintenance and cleansing, right? Which is a very very equally critical part to it. So wherever you would like to add some color to it now that you're here Yes. Yeah, Arabia short intro, I think should have been done. So son by Suzanne, but since that I asked a question, you can do a short intro of yourself, and then go ahead and answer this question of shrinivas.
Okay, so, I am Surabhi. And so to tell you a little about my background I from an entirely different field and A scientist 15 years after my PhD and that is in the biotech field and I started see coming in from the IML field into the biotechnology area, and I was very interested and I was looking to step aside from benchmark. So I felt that this would be ideal. It was a great risk that I did not know coding and I had was from my class too. So it was a great risk but what really triggered my decision is that I was seen plenty of AI ml, this research are sealed and they were leading to great applications. And so, I I researched a lot on which program I should join to learn in the best possible manner and when I compared all of them the best one that I felt at that point was talentsprint and ripple it Hyderabad based program for multiple reasons it was it was a practitioners curriculum and it was giving us a head start and a lot of pre reading materials and you know, pre fundamentals. So, even before the parliament is I needed some preparation, so that preparation material etc I got completely from them, and that helped me a lot and Yes, sir Basically, my area was drug discovery and diagnostics. And so I, I intend to start something of my own soon and in the interim, I'm learning more and contributing here at talentsprint. So coming to So, Srinivas was can you just repeat what you were talking about?
It's talking about maybe she can repeat the question.
I don't want to be integrating indicators. Oh, hello, survey.
Yeah. Nice to have you.
So, my question is like, we help train or train the models on toy data set until now. Yeah, right. So when we when we came into the this, companies like to attend data sets They have less number of data, data set data samples after the ETL process.
Yeah, on the less number of data,
data samples, how we achieve the metrics?
Yes. So, you will gather, like anyone, any participant and will gather huge insight in the area of using real world data sets. So at the very beginning, the first so we have four modules, okay. And it starts with fundamentals, and it goes into algorithms and a deeper look into algorithms deep learning. So it goes hand in hand with real world data set based on many hackathons and hackathons. In this you are completely dedicating your time to solving problems based on real world data sets. Okay, so now Not only a systematic procedure of data munging data cleaning, and all of these very essential parts, you know, an entire hackathon is based just on this part how to deal with real world datasets. Okay, so and the performance metrics, there is an entire section dedicated to just learning when to use which performance metrics and how best you can interpret your data and your analytics based on which metric and all of these things. So, you know, a lot of online courses will take you through concepts, but nothing in depth, okay? And nothing like a live session, a live session, you will get to ask your questions to the professor. Right, and the professor there and then He has answers for you and the professor insert for herself is working. They are researching like they have the lives are dedicated to research and they are the they know the exact information and everything.
Okay, thanks for ABC, I think the short answer is actual solution for this syniverse would be a complete consulting project for a specific company. But what happens during this program is to ensure that you are exposed to the implementation challenges and how to go about it and more importantly, identifying the process of data cleansing and data gathering, which will make things better because end of the day, your domain data has to be created in the job and that's something nobody can do it for us. Right. So I think that's probably the answer. For your question, and in the meanwhile, I think Sunil his question in the chat, Sunil Kumar question has been answered by syrupy because what we do is I think it's very important that somebody who starts the question is how to start a data set without knowledge of pudding. You don't need to start with coding knowledge. But definitely you need to develop coding knowledge. Right? And a little bit of mathematics knowledge without this you can't, one can't expect to be. I mean, this reminds me of a very interesting question. I used to go to campuses and ask people, how many of you want to go to a seminar of thousand people? If I ask a question, how many of you want to work in IT industry? 99% will show that MCs, right? How many of you love coding? Only 5% of the hands go up? Right? I think that situations for a ml can't work, I'm sure robotic process automation and rule based engines can work for that. But for something like AML I think Understanding both how data has to be analyzed the far the algorithms and also a little bit of coding, you will not be doing much coding but I think four lines of code will take a lot of time to write, but those four lines will be extremely important and critical for someone to proceed. So what I will do is now I have taken to three questions, I will go back to the deck, I will probably show two three slides and come back to discussions after that. Everything is numbered one, I think, probably a typo but I'm sure all are equally important. So, clearly what we have seen as a program for working professional needs to be very different from what is done for
academic students. Right. The use cases are very well known to the professionals. They have been working day in and day out solving real world problems on the patient's within quotes. I recall it of going From the statistical concepts to coding concepts to algorithms, and how the algorithm works, and then looking at the solution never works for a professional, I mean, forget professional today even for students, it doesn't work in many cases. I give a very simple example right? When I was younger I was I mean, I used to live in a cage fan and they play cricket. Right. And I did not know what is which is covered which is, which is clearly when a data shot somebody said you did a very good cover try then one day I realized that it's covered. But I used it to solve a problem of scoring. Right? I am sure when it comes to work, I have done many things. And I'm sure each one of you have done many things. And later we get to know what it is and develop from there. Right. I think it's a practice theory deeper practice is the theory is the concept in which Professionals programs should be worked on. And that's something that AAA at Hyderabad have clearly understood. And that program for working professionals is varied from what we do for their own to their own graduate students where they started statistics analytics, looking at mathematical modeling, and then get into algorithms. Right? So when we really look at and when you look at the results of the survey, clearly, question number one and question number two specifically represent number two, starting with problem definition of solving the problem. And using algorithm for it and getting to know how the algorithm works and what's the concept behind it. So that tomorrow when things change, we are able to adapt this knowledge to something else is what professionals are looking for. And that's how the program is designed. If you really look at it, the module zero what we call it Is foundations to get everybody in the same space? Because somebody with 15 years experience somebody with three years experience, how get into the program with different levels, somebody will be very comfortable in coding because they are that person scoring today, and somebody did it 10 years back, but understand the business case, far better. Both of them are brought together into the program to the module zero. But module one is defining problems for solving using a algorithms. It doesn't start with what is classification. It doesn't start where it started. What is recommendation engine? It doesn't start with even the mathematical concepts of macro system algebra, right? It starts with understanding the real world problem, defining it for the machine learning to work. Right. So that's how the program is defined. And that's probably what makes this program very different. I guess I said 2300 people, we are so far done 14 cohorts. 2300 people have gone through the program and the Philippines cohort is being launched. The second one is more hands on and group work. I don't think any executive today at this point of time working individually, right, we all work as groups. Somebody takes some part of the job, someone else takes some other part of the job. We work as a group and in this presentation, I think the numbers work together to make it up and show it to you. So I think group work is what executives are comfortable with. And the program ensures that it's hands on and group work that is done. Right. And the third part is, lab exercises need to suit the specific need. One of the questions that was asked in the poll, if you revisit the poll results. Let me just show share results again, was
professional professionals are different levels require the experience level will have different objectives. We I personally know a lot of people who are at the director level or senior director level or president level joining. And they say I am extremely happy to know what happens and how the algorithm works. Right? I have a team, which will make the algorithm work in my team. I don't want to get into coding beyond a point just to ensure that I really want to know how sensitive is this algorithm to changing in the data, how sensitive is this algorithm to something which is changing in the in the real data set? Right? The other category say, Okay, I am in the middle management level, I want to look at how the code works. I want to be there. But I also want to be in the senior level because I want to evolve into use cases more than anything else. That's the second pillar. And the third level is I have a quarter three years of experience. I want to make it work. Yes, I understand domain a little bit but I am still not there. What you In this program we do is we create the same exercise, which people can do it in three different levels. Right and make things for the fourth point of it. Now let's stop sharing the results. Let's go back to the deck. Let's go back to the deck. I think peer group learning, that's one of the biggest things that we have realized in this program. Right? I think seven eight startups have come when people from different companies and different domains have come together to solve one area, right? I know when agricultural startup came with a person from biotech industry, a person from the idea that we find a person from the retail industry coming together and starting something right. Effective peer group is extremely valuable. I'm sure most institutions are made with such peer groups. Okay, so that's the background. And what I will now do is take three four slides about the peer loops that I talked about and then how we will learn and then take more questions. The error program this was launched in Jan 2018 2300 people over 14 cohorts are completed the program of 13 have completed another hundred and 50 people are going through each cohort is around 150 people. And the profile of participants is like this, these 2300 people are from 740 companies, there is a 70% of the people have five years of experience 19% demand 30% from startup, these two are what we are very proud of 96 97% completion 96.4 or 96.5% completion. And under 3.5, I'm sure a lot of reasons of political and other related reasons. So typically a person who starts the program completes the program and the net promoter score. I mean, this is something that would be of interest to you. This is what people believe at the end of the program, saying that whether this program has helped them and will they get their closest friend related to join the program and the score is taken this way. Definitely Yes. Is x percentage, maybe is y percentage, and definitely not as or may not, is the third category. So if there are for example, Santa Fe percent people saying yes, definitely yes 25% people saying maybe and zero percent saying no 75% minus zero is a NPS score. Typically, any NPS above 30 is considered good, and any NPS above 16 is considered great. And we are proud to get 75 thanks to the participants. I'm sure the learning has happened to faculty and talents deep but I think peers also made a huge impact in making this happen. There is some error here. Okay, so column one, column two, it says
p 70% of people are about five years experience those who join this program 80% of the people that 42 or more than 10% 2% more than 10 and five to 10 years is 28%, three to five years 18% and less than three is 12%. Right? This is how the professionals are different who's comfortable that these are the companies of course, definitely not possible to learn. Read it. It's very small on certified portfolio on 50 companies, but these are all companies you can think of every company the Microsoft, Google, Amazon, Infosys, Wipro, TCS, CTS Capgemini. HCl then the top bangs top retail companies, people from all these companies have joined this program so far. This is a very interesting slide. In generally this ratio remained across all cohorts. There is small movement here and there. One third of the people are developers, one fourth or another's one fourth or managers and almost one eight is leaders join this program and when they work together in groups The learning is not only coming from the faculty not only coming from the mentors in the lab will give one on one and group support, but also from one another, which has generally made this program highly effective and efficient. So that's with that my first flights are over, I can open, open up for questions. Any questions, anybody? You can either type or show your hands and ask. I'm happy to take it in. Yes, I have. Karen as multitask.
You can unmute yourself, Karen and please talk
to your audio Please go ahead.
Yeah, sorry. Can you guess
I'm sorry, you
can't you can unmute yourself and speak. I think there is some disturbance because of which we are not able to hear you can repeat it, please. Yeah. Okay. So I seen your question on the chat. This happens on weekends. Right. This is a program that you are expected to spend Saturday, Sunday a part of the day. On this program. One day will generally be lectures, other day will be labs, right. And you will go through the labs, along with mentors and the lectures will be from top professors, weekend Sunday, so that your weekdays are not disturbed. And after every three, four weeks, there will be hackathons and stuff like that, where you will work as groups and come up with some real solutions using AI algorithms. It's a really hands on program when you really look at it, I think more than two thirds of the period you will be doing things and so are many of us asking a question about technical expertise, the expectation is that you should be able to code and if you don't have it, we do have some people who joined the program saying that I don't know how to code. I don't really want to go but I want to know the outcome. But getting certification and terms of assessments and not accessing those may not be different, difficult, maybe difficult. So it is about there is a module zero which will get you ready for coding, right there'll be a refresher, that's done. That could be a four five session during which you will get the complete refresher and you will be ready for coding. What is required for this program? It will not be heavyduty need of coding, but you still need coding. So does this program find your customers for different segments? See, the program is called In the labs are customized the program everybody needs to understand the algorithms and how it works and God what algorithm can be used, but the positioning of it or the labs is bad, there is a difference. Individual labs, we will have three difficulty levels are three levels you can choose from, and anybody can choose any levels. While somebody wants to know Yes, I have learned this algorithm, this algorithm is not going to be too much that I'm going to use, but I just only want to see how it works. So that's one category. The second category is I want just not just want to know how it works. I also want to implement it and CH, right. The third category is I will write all the code implement and make it work. Right. So you as an individual participant can choose between any of these three for any of the labs. We are not forcing saying that if you're a CXO you should only take level three or your refresher and coding Quarter developer you should only take level one. It's not like that anybody can choose any of this and we always see that our average people take second difficulty level they will do 20% of the things in level level one and level three right. Ah okay would you even share the syllabus and detail learning part okay how many months is program with the faculty a lot of questions, but let me take some of them. The the
primary mission learning related lectures will be by total it professors and T top research faculty from triple it, the coding the related labs and other things would be by mentors who are from either industry or are have gone through the program or they may be a triple it students or a talentsprint faculty. The third category is group labs. You will have to do on your own based on what you have learned as a participant in groups, and hackathons. So that's how the structure is done. How many months is the program? What is the fee structure? So this program has been running for last two and a half years. The program fee is around two lakhs plus GST. There are special scholarship available because of COVID. Currently, on lockdown, there are specific scholarships available for people to join and also special schemes on that, which our team can definitely help you out. And you get a 24 month zero percent ami, given the current situation, I'm sure people are not a good while. I think this is a very interesting time. I mean, I used to be in the financial services sector. I used to keep giving one example, right, I was an insurance sales and as you would know, insurance is one of the most difficult things across the globe. And I used to joke around saying that the best place to sell health insurance is outside Apollo hospitals. And that's The only person who are not eligible for an insurance, right? So it is like the same situation today we are in, I think that companies are going to automate, I am sure. One of the surveys we did recently I think 80% of the employees in IT companies believe that 20% of the jobs will go, there will be pink slips etc. and learning deep tech is going to be extremely critical. And this is also the time when cash flows are not very easy. There are salary cuts for many people that are forced holidays and forced. What do you call the witch creation and people are asked to go on holiday for six months and come back. So what we have done is to make the payment process simple and easier. With a 24 month zero percent EMI, I'm sure those things on our team would be helping you out with this. There's somebody who asked a very interesting question which I would like to take that question He's the remote training effective. So let's put it this way, I would like to I should very transparently share what has happened with this program. As I said 13 cohorts of the program has gone through so far. Almost the first 910 programs and 10 cohorts of the program happened with 14% to 15%, physical and 60% 50 to 60%. Online. When I say 40 to 50%, physical, the lectures 70% of the lectures happened online. And sorry, 70% of that lectures happen physically 30% online, and it's the reverse for labs, right. 30% of the labs happened physically 70% online happened after the COVID and even before that we had lost already a four by four people from across the country joined the program. And the whole structure was done in such a way that people come for two visits campus, visit Otherwise they learn completely online and build the COVID structure to be precise, cohort 12. Module zero and module one were delivered physically and module two or three, trade completely online, right and more of the 13th cohort and the 14th cohort went completely online.
While we also had the same
concern whether it can be effective, the good part is the NPS which are 65 70% before went up to 80 85%. During this period, of course, we went out of the normal process to ensure that there is more one on one support, there is more support time available for people, the faculty is able to spend more time and hence, it better serve the cohort 13 and 14 more effective or as effective as the usual cohorts. No, at this point of time, I can call frequently say that our format of online I will call the friendly everyday, we will only do interactive online classes. Right? When the lectures are scheduled for one and a half hours or two hours, it happens with the faculty in front. Only pre learning is recorded videos pre learning is PDFs, but actual lectures are delivered synchronously. Actual labs are delivered synchronously, where the mentor is available on hand at that time when you're supposed to do and if you want to do it on your time, you can fix a one on one mentoring specifically specific slot with a mentor. So these things when we did we find that the online program is as effective if not more effective than the classroom program, because some of the problems of traveling and stuff like that has been driven people are able to take classes and things like that. And I think it's it's very clear we are moving towards digital era. And I think coming back to the social distancing is here to stay and coming back to the classroom with 150 200 people is going to be a little too far away than what it is today. And when that happens, I'm sure we will be happy to invite all of you together into the classroom and take the session. Ah, there are there's a question from Narendra asking about cloud environment, we have everything in our LMS. The ipad.ai is the platform that we use. When you log in to your lab exercises and hackathons are integrated. If your exercise requires very few require it, a virtual machine, you can log into the virtual machine from the learning platform, or Google collab is one of the tools that we use, you can go into Google collab and complete your exercises. GPU and other things can be used there. Right. So those things are taken care of yogini is asking for the syllabus syllabus is there on the site and there is a detailed FAQ on that you can go through it in case you're not able to find it. You can please write to us at ml at talentsprint calm we will send you the detailed syllabus for you to go through. Yes yogini. We also use Python Python for the program and that's that's probably the most widely used programming language for AML and the match background that survive kayvon is asking yes match background is required. See, the background is we are not going to research that this program is not for people doing the such. Those programs are different that we did four years six year programs. That person can do this. program is for practitioners. And if you are a practitioners and want to build AI applications using the algorithms, that's something that you will be able to do, right? What kind of algorithm to be used for which particular problem statement that you're going to solve. And for this, the math required is you should not fear the terms of mathematics, you should understand what probability is, you should reasonably understand what a matrix is what happens when you do a matrix multiplication and should not see what happens when you hear a word which you are not comfortable with this subsequent sentences that the faculty is delivering. Right? It probably your mind will stop when the difficult word comes in. That's happens to everybody not just you and you it's also for me, right? So familiarizing with mathematics. It's very important. So, we would like to add color to this.
So we are aware. Yes, please can you Oh
yeah, in my case also, um, as I said, coating and math, both of these required some preparation before the classes, but that material was available from this course. So before the classes started, there was a 15 day period. And I myself started preparing one and a half months before the course, but a 15 day material like one our each day if you give and if each of so our math knowledge or math knowledge is usually a bit rusted at this point in our careers, but what it does is it just gives us the division we need and the exact concepts That will be used in the IMF later on. So that really helps in bridging the gap. Okay, so I don't see anybody on that account. I have myself gone through it. And it has been very useful. The materials so when you join this program,
how many years of gap from your previous getting into class and this
gap from what?
From the education that you had? Because that's at the time, right?
Yeah. So I lost that. I did. Like anything, really math was during my plus two, I mean, 12 class, and after that, ah, yeah, that would be 20 years back. So after that, do then I was a scientist. You research whatever math was me. Did it was not what we usually are focusing on here. So it's very different. But the exactly 11th and 12th class is labor as a lot of that may be useful because I had a good foundation in that and that really helped. So I'm sure each one of you if you are up to that level, and anyways, you must be using a bit of I mean, teaching your children also and so, I mean, I don't see a concern on that aspect because I have gone through it. And believe me, believe me, it was a lot of apprehension. I, I mean, I would be the last year at this time, I was the most apprehensive on any account.
Sure, so there are a few more questions. I will take it up When the Fed faculty manage hundred people into our session, when the classroom happened, it used to be 200 people. And any top course in any institution happens at even 602,000 people. Right? The interesting part is, I think the faculty who are experienced in handling such sighs or the faculty who are able to understand the questions and are able to get their together, delivery ready for that, and secundus in the online mode, it's probably far more easier and effective to get their questions answered. I was just there in the last day of class. That hundred and 40 150 people are attending 60% of their queries via chat that's sorted out by other students. Make the faculty not even have to take it. And the best part is the most important questions and most difficult questions goes to the faculty and this area and choose that The peer learning is far better and the faculty staff is being effectively used for the right things. Even addition to this, you will have forums in which you can post questions and the mentors can answer you. Other students can other participants can answer you. And there are specific doctoring sessions created separate slots in which Pacific doll clearings are done for every one every module. And in addition to that, there is a process of one on one q&a that you can fix a 15 minute 30 minute slot with your mentor and get your doubts clear. And the one thing that we have realized as talentsprint when we have done so many programs with top institutions like IAC, I am Calcutta triple it IIT Kanpur it has a bar. I think the technology is enabling us to do things better, faster, more efficient. Yes, there are some shortcomings like group work and knowing each other, which are now being sorted out, thanks to group work and the breakout rooms and other such solutions coming in. Right? I think it's it's working pretty fine. And I should be saying this when we did the online cohort, the 93 was the NPS. And just to tell you, the global top, I mean, I don't want to measure it individually one now one particular cohort we like to look at overall, that's why I talked about 75 and others. But just to tell you, there is only one company in the world, which has more than 90 NPS. And our last online cohort 100% online cohort had 93. Right? You can search online, which company I'm talking about. Right? And so clearly they both the effectiveness and the comfort and more importantly, the ability to learn. such deep tech from a top institution like Lady Hyderabad within their certification, propriety certification and alumni membership is extremely useful for professionals in today's context where deep tech is going to be the future because I think NASSCOM say 75% of the applications Indian software companies developed will have some component of Yeah, and 1.4 lakh professionals are required that expertise while institutions are churning out less than few thousands, right. So with that, he can do this. Ah, okay, there are a few more questions what software or inbuilt technologies used to develop machine learning algorithm Dr. Arthur Kirby see our program is done and most developers today are using sci fi and related Python based pandas kind of areas. So the program will help you to get hands on with that. The lab infrastructure required as part of the platform that you will be giving and getting access to the new driver for this program and from that, you will be able to take it forward. Matt was asking about time space between us Canada and India Institute challenging now, probably I would suggest that you can we can discuss individually on this, we will find out what the structure is. I think the only challenge will happen. See the lab Sessions is easy to manage and help you out. Even though us Canada timings. The lectures from triplicity faculty will happen in the specified times in the Indian ISD.
category. We need to work this out. There are a few students who are participants who have enrolled for programs from us. My team will be happy to assist you. You can reach out to Susan right Susan. What is your If you can share your email idea, Suzanne will be able to handle you and take you through on how to get this done. Are Dr. nagaraju is asking a very interesting question. I wish my colleague ashokan was here to answer this. See, for some reason, the researchers have used Python extensively, not just not not just the researchers in computer science, but researchers in aerospace researchers. In T, medical health care, all researchers have adopted quite often it is I mean, I have learned other languages earlier and I got exposure to pattern after when we launched this programs. I think it's intuitive language, language which one can explain in the normal terms and lots of algorithms I'll be developed using Python. And that's becoming an open source is being the I think the whole aim of the world isn't open source. And that way Python is probably ruling the roost because it's one of the most widely adopted language. I mean, even 10 years back, I realized that US universities, top universities used Python as the first learning language, because it was very logical thing to do. And the syntax was not coming in the way and it was helpful for people to adapt the language. So if you would like to add something to this,
Yes, I can. Also it has incredible community support. So a lot of explanatory material etc is available for Python. It is simple and easy to understand and learn and a lot of packages like pandas, NumPy, and psychic learn These made Python are very good choice for machine learning activity. So, and yeah, so in general, they are preferred by con, but there's no. Like over the years Python has gained an advantage maybe because of these reasons.
Not Prashant, you have asked to speak you wanted to speak I have a hunger. Can you please go ahead, unmute yourself and ask your question.
Michelle, can you unmute yourself and ask a question, please?
You're on mute. Yeah, you want me to bring
in extra brief about the difference between machine learning and deep learning
Sort of we'd like to take it up.
Yeah. So, you know, if you think of AI is
baked in as a field
the any computer system that mimics intelligent human behavior or if you just want to summarize in one line, a subset of that is machine learning and a subset of that is deep learning, okay. So, machine learning as you know, it will use a lot of examples to learn from the data analyzed. And within the machine learning domain there is deep learning, deep learning is nothing It is. It is based on neural networks and neural networks are You know, these are like connected information processes unit that are connected. And I mean, in short, I would just like to say that deep learning is the reason why maximum applications have been coming in the past few years. And it offers tremendous potential and yeah, so that is how these are related to each other these fields. Yeah.
Thanks. I think there is a poll that has been lost by the team, please answer them in the meanwhile we can keep discussing various questions. I have a couple of other people I have around who was around, perhaps.
Yes, no, please go ahead.
Um, you're on the mic, please go ahead. Ask your question. Please.
Can you hear us? Yes.
Yeah. Can you hear me?
Yeah. Yeah. Good afternoon all. So my question he means some time before that question is raised for that, the ML versus that deep learning. So all that come all that scenario of a melon also available in deep, deep learning also. So in what scenario we We'll choose deep learning what and in what scenario he will go for ml
okay. So, let me let me take it Do you want me to respond?
Okay. So, as I get the question you're saying in what way in different scenarios ml and deep learning. So, um ml is addressing a lot of the I think the analytical and so these are different types of algorithms, okay. For example, deep learning algorithms, these are very adept at doing certain tasks, like image recognition are Speech recognition, things like that, that that very complex data when it is very complex data, deep learning provides better accuracy. And yeah, so that is why deep learning already requires a huge amount of data. But it always it gives a lot
better accuracy, etc.
Yeah. Thanks. Thanks. Sorry. I think you're already seven minutes beyond our time. I think one question, I think sort of see now wants to speak as requesting you to unmute yourself. Go ahead.
Hey, hi. Can you hear me?
Okay, so yeah, so unfortunately, I've registered for this ama course through Great Lakes but I thought I'll join this course to understand more and like gain more deep understanding and this so my question is, I heard some recently off of data scientists in the recent times and it got me thinking that are the rules of data science and ml in the companies right now is something which the companies are experimenting experimenting with, or it is with long term focus, like what honest advice you can give to take it over the current expertise of people? Or is it like a learning which you should keep along with like, you know, a language you should know this as well. My question would be like, are things like auto ml going to change the data science future?
So, okay, so let me put it this way, two years back when this program was launched, and today, when I look at the curriculum itself, it has changed at least 60% of the development in the machine learning field is so fast, right? And I think the turnaround time from the research lab to the production is extremely fast in today's context. Right. But the one thing is very clear, I was just starting spending some time with Professor Jehan and I know some people it has a body earlier this month, right i mean the last month I jumped right on it clearly, the basic concepts have always remained the same. The algorithms and how things work have the membership, it is the type of data, the power to process the data, and the algorithms which are able to exploit these things have changed. So, let me just start with the first question. Is it too early? Is it something that is going to come tomorrow I should learn today? The answer is no. I think it's already arrived. Right. One of the most recent programs is something that we launched with I am Calcutta called a poverty marketing. Right. The other program with Indian Institute of Science is approved I'm on digital health, which is how a is helpful in healthcare and imaging, because imaging is the top part of healthcare and all the X rays and ECG reports and stuff are all on image basis. Right? So aryaman and data science I would say are the two sides of the same coin right? one looks at the solution from the data point of view, the other look sort of the algorithm point of view, right one is the hardware and the software if you would put it in a different context right, the real data which is hardware and algorithms, which is a software which helps unlock the value in the data is the algorithm. So it is today being used extensively, not only by top companies, like for example, if you have it be extensively used in our education programs, use a extensively right? We use it for people sitting at home and taking an exam. nation property them. Right? I exactly know who is at which point of time taking a support from somebody who he or he or she's not supposed to take. And I can go and unlock on top of that person's test or processing the video that is generated in this two hour session of classes and making it searchable for students. So, every institution, every corporation, every company is using machine learning at different levels, deep learning at different levels already. And what coding was 15 years back probably, machine learning will be today. It is necessary for people to learn this and be available for I mean, be ready for available opportunities which are coming up because India works extensively in the altos work from across the globe right into Development right. So, for that it is extremely critical. And the second partners you also heard that various education initiatives in India are being taken to inch include a machine learning or D school at the college case itself. Right, by years from now, six years from now, as much as we have number of engineers, we may have a number of UI ready people, right. The question is next five to 10 years, when the I mean, I would already I would say that the waiver started two years back probably another five to eight years. It is there for people are already working to be ready for getting India into the global map in terms of AI competitiveness, right and that projects are coming to India. The people are making do with certain things program certain trainings and getting. I mean, I am sure there are some people who are practicing machine learning in this seminar itself, but they may not understand the algorithms and how and why it works. It is critical that this expertise is built for people who have three to 10 years of experience or even 15 years of experience. As the changes happening faster today, and the supply from the education is going to take five years more to get to then we have to be the torchbearers, the working professionals are these ambassadors. I guide the new generation that will come fully ready from the college system.
I hope that answers the question. I think if there is nothing else, I hope all of you have taken the survey. And I would like to thank you for being here in this session. I have I think, some people who have not given feedback please go ahead and share your feedback. And thanks for joining me today. Thanks, everybody for joining us. And so, Sonia, anything that you want to add? Thank you. What is the next chord starting so the next cohort would start from August. In case you want to know any more information regarding the program, the details of the team, as mentioned in the chat, you can get in touch with any of us to know more about the program. We'll be glad to help you out. Thank you. Thanks a lot. See you See you soon in the class. Bye. Bye.
Bye, everyone. Thanks. Thank you.
Watch the entire interview here https://youtu.be/CO40wIOz-sU