Hands-on AI Learning with Vocareum AI Notebook

Good morning, everyone. I'm Sanjay Shua. I'm founder and CEO of Ocariam. So thanks for coming. So this is great. Didn't know what to expect since the, in the last day of the conference.

So first of all, how many of you, just if I can get a raise of hands, show of hands of how many of you have heard about Bocariam? Or have no idea who okay. So you'd want, two. Okay. Okay. So, So in terms of the presentation, I know the I'm assuming the most recent, the biggest reason why you guys here was to talk about what we do for the AI.

Right? But there are few slides. Hopefully, you can bear with me, which will talk about Vocaadium in general and what we have been doing in the education space. And, I was not sure about it, but now I have been able to use my own laptop. So, actually, I can do a demo just to do a quick demo. So the as as far as the product is is concerned because, you know, it's so much easier to show it than talk about it.

So just to give you a little bit of history about the company, I started this in twenty fourteen. And, and someone was asking me what, what's the name Vocadian, where it has come from. So It was kind of made up from Vokare, which is like, you know, on Vocation. So the idea was, hey, how do we teach? What can we do to contribute for people learning skills, which can help them get a job. So about ten years ago, twenty fourteen, everyone was talking about you know, computing skills.

It's like how computing was becoming foundational and how you need to teach, you know, in high schools, And so there's a lot of focus going on into making, you know, a lot of legislation being passed to say that now, you know, computing can qualify for your graduating requirements and APCS and APCS. And then in colleges, you know, a lot of colleges are saying, okay, how do we make sure that everyone learns compute, not just the computer science degree. So that was kind of a lot of focus for the company. But as we started sort of engaging with the market, what was really fun to see. And that was, you know, it was it has been a long time since everyone was talking about, okay, computing a new foundational skill we needed with math, you know, we need to do with language.

But if you start roll the clock forward ten years, And things have only accelerated. You know, the big thing we started hearing about is, of course, the data literacy that everyone, you know, needs to become, you know, at least to be you know, comfortable with using data for, you know, making decisions and making for for getting insights. And of course, what happened the last six months has been really exciting, you know, it's about everything about AI because at least, you know, we believe now AI is has to be a foundational skill. Not that everyone has to be able to program a model, but you have to be able to know the difference between between you know, difference between being AI literate and not solar trade would be that you would not know what's the difference between different models is, what the models and you have to use the models for productivity, but then you have to know that you can't, you know, when it starts hallucinating. And how can you take a model, you can pick foundation model, and then you can add your own data set to it to, you know, make your own model, right? Maybe it's a proprietary data set.

Maybe it's very specific to, let's say, health care, I know someone was about from a health care field or whatever. So unlike what other companies are doing on AI, we have a slightly different focus So we are not using, AI to, you know, improve our product. We are not even using AI to help you improve your teaching. But we are actually, delivering AI labs to help you teach AI, right? So The idea was that, you know, as you start, teaching all these foundational skills like computer or data science or now AI, One of the things which you believe that the learners need and the educators need is really easy way to use new tools. These tools cannot be installed on your laptop, often it needs access, via cloud.

And these tools Like, you know, the stories you hear about students, they get excited about learning computing or data science, and the first thing they're asked to do is to install a whole bunch of tools and this takes, you know, them three weeks and there goes that all this excitement about learning, you know, So that the tools need to be accessible to the educators, to be accessible to the learners, and they need to be accessible in a way in the context of education, which means you have to be able to do assessment and you have to be able to, go out there, do team projects and collaboration, and we'd be able to give exams. So anyway, that's kind of a little bit of the context. So about, what we're gonna talk about. So just talked about our mission is to deploy and manage virtual, and the product that we call is virtual labs, right, for, for digital skills. And so here is like we have encapsulated.

Here are sort of, all of our products, and I will spend most of the time on the AI, but, because that's the latest greatest product. But here are the products that we have been delivering, so for the last decade, Vukaryam notebook for Learning, AI. Then there's a Jupyter notebook for data science literacy. This is again very popular, full stack programming environment for with VS code for if you're learning, computing, if you're learning how to build applications, VOCadium's cloud lab. This is probably the biggest deployment.

So we are there in six thousand schools out there because of a partnership with Amazon. So if you have to learn mean cloud computing has been one of those skills. Maybe maybe not so much foundational, but definitely have been great for, from an employment perspective. Right? Because the ten million IT people need to be teaching, you know, to be educated on cloud. And if you go and get a cloud certification, you might get like hundred sixty five thousand dollars or whatever the average salary is some some ridiculously high number.

So we have that has been pretty exciting because we have been able to scale it out We have about three million learners out there worldwide. Then we just launched a similar partnership with Databricks. So So data bricks for those of you might not be familiar. There are actually they were one of the they and Snowflake were two big winners from the whole data perspective. Like, you know, now you have like massive amount of data.

It comes from all these sources. And you have to store it, and then you have to process it. You have to process it for getting inside, or you have to you have to story it for driving your, maybe maybe machine learning. So data bricks just partner with us to say, okay, then we are going to be providing learning. Once again, it is from a jobs perspective, it's a big deal.

And the last one is a cyber range. So a lot of you have talked to in the conference, have been talking about, hey, I'm launching a cyber security program, and how do I how do I give you can't teach cyber security without a lab. Without actually doing a hands on work, you can't just, you know, the, the presentation will only get you so far, you know, talking about, like, you know, How do you identify attacks, how do you identify vulnerability? Again, it's, it's being taught at a large scale and we are participating in it. So AI notebook. So we will talk in a single talk about a little bit more and give you a demo on it, but the whole idea is that you can teach AI at different levels.

One is maybe you're teaching prompt engineering, how to how to send, you know, prompts and also assess the prompt. So you have to be a student have to not only be able to write a prompt, but you being able to have to be able to submit it, grade it, and you have to be able to send a prompt to maybe different models. Maybe it's a chat GPT, maybe it's a, Google's new one. I guess, Jeanie, they are working on it or, Lama II from Facebook. And you're talking about mid journey, right? So how do you And then the other aspect of AI is like fine tuning, which is to be able to take an existing model and maybe add your own information, maybe add your PDF or whatever and, you know, create a new model.

And the third part is building, you know, sort of real foundational, deep foundational skill about how do I build a model from scratch, right? So those are kind of three things. And depending on the context, this is what we are hoping to deliver the laps. For. For data science, as I was mentioning that we have, fairly large deployment of a Jupiter notebook, Jupiter notebooks, for those of you are not familiar it really, made programming accessible to a lot of people because you get, you know, it's really nice. You don't have to learn you first want to start coding, you don't have to start worrying about learning Linux or, you know, you can just write there you write a piece of code and you say run, and it gives you results right there.

You can drop instructions. So it's made it. So essentially, which means anyone who's using computing especially our business schools, you know, it's a big deal. Like, so there are a lot of business schools which are, which are incorporating computing, but everyone there is not a, you know, Linux part of whatever. So that has become the, way to, engage with computing, right? So There are all these packages available.

So we did the four lines of code. You can get a nice insight. You can get a graph and chart, and you can you can play with lots of data. Full stack programming. So this is like, you know, your hardcore programming.

If you're learning programming, the VS code has become the go to environment which we support, and you can not only run- run programming, but you can do full stack development, so you can do back end and, you know, and, and the front end development for, for deploying applications. Cloudlabs, as I was mentioning, this is for you, of, this is this is to learn foundational, you know, cloud skills. What is exciting about this is from an employment perspective is that AWS Azure GCP, they all have certificates. So when I was saying six thousand universities, a lot of it is actually being deployed where the students are not only learning cloud computing skills and getting credit, but also they are being prepared to go get the industry certification. Could be AWS, and, you know, and that certificate definitely, you know, adds a huge amount of value in terms of being able to get an employment.

Data engineering as I was saying this once again from a from a employment perspective, this is, again, is going to be a huge skill. So not, you know, so when we say data engineering versus data analysis, so the idea about data engineering is that, you need a lot of people to take all of your various data sources where you're getting from maybe a point of sales or maybe from your log files or from your, maybe spreadsheet, you get all this data and you get, you know, get it ready for the further downstream data scientists or, for analysts. Right? So there's, so we are pretty excited about this partnership we expect again, once again, to reach millions of learners to learn data engineering skills on, databricks. And one of the questions always come up when talking to educators. Is like, you know, how important is a given tool? Right? You want to teach data engineering and and I don't have the answer and I would not pretend to, you know, want to get in the middle of it.

But clearly, you know, if you're learning data engineering and if you're learning with the real world, you know, tool, it's better than art, right, but that you need it or not. And of course, one of the one of the challenges, the tools are always kind of changing. So you can't be chasing the They test your test tool all the time, but but if we there's the gap between so I come from semiconductor engineering background before I start a company in education. And for those of you guys know, one of the big problems since semiconductor teaching it in universities is that, you always got like one or two generation old, you know, data, like for process notes or whatever, because everyone was really scared of it, which means, and that one or two generation, in the earlier days of more smart meant you were like maybe your skills were off by like a couple of years or three years. Now as Moore's law has slowed down, it means now your skills might be off by a decade.

You're coming out there and you are actually in university. You're learning something which is, you know, so that's part of our mission is to, like, you know, especially with public cloud to go to all these companies and say, look, we need to go out there for our, you know, student. We need to provide them, you know, latest greatest tools. Sometimes it's, you know, Some of this conversation are working better than the others, but we are making progress, and data bricks is a good example. Cyber ranges, as I was mentioning earlier on that the cyber ranges, is becoming a big deal, because lots and lots of cybersecurity jobs are being, created every every day and they're just not enough qualified people.

So you can launch a cyber range, an invokerium. So you can configure it. You can say, okay, here is my three machines. Here's the vulnerable machine and you know, here's a machine which is going to use it to attack this machine, and you give students in a project to go and identify it. And so this is, So every learner can get their own, they can get their own personal cyber range.

And if they make a mistake, they mess it up, they just hit reset, and we just tear it down and start all over again or whatever, right? So you can go, you know, work with it So, I mean, that once again becomes a big deal about cyber equation. So I know only two of you raised your hand, so you know vacarium, but we are there in, about six thousand schools. There are a lot of times because of not directly, we don't have a direct ship book, but we're a partner for someone like AWS or databricks or whatever. There is, sometimes, Some of the big partners here is Georgia Tech is a big partner and, Purdue is a big partner, University of Washington, Notre Dame, USC, USC. Columbia.

And, for us, this has been pretty exciting the whole canvas integration, right, because it has, for whatever reason, not only in our academic, not only our universities, but even our professional, customers, and we are just seeing Canvas, you know, being a dot of everywhere. So it has really enabled us to really invest in doing a really, you know, deep, as deep integration as possible. And, what it means for us is, obviously, LTI one point three. And once and again, it goes back to the earlier point of students should be able to get, so people get excited about, like, saying canvas pages. You can be giving them instruction.

It can drop a little lab. You can say, okay, you know, you can drop any lab. You can say, okay, do a little bit of work and check your work and go on or you can do it in assignments using LTI. It's a second part. I don't completely understand it because but I see lots of conversation happening is like you have a course and then I need to I need to distribute it sometime to a few hundred faculty because of the central, you know, content being created.

And so how do you manage all that? So we spend a lot of energy worrying about LTI links and how do they move from if you want to rerun a course or you want to duplicate a course or you want to do a clone by reference. So you change one place and it changes everywhere. So there's a lot of energy goes on into the, you know, into Canvas. How do you sort of just manage content? Grades is an obvious one, of course, you know, so there's a lot of assessment capability built into Vercarium, which is, I just want to be cognizant of the time so I can go down to the AI and realize that. I didn't want to talk a little bit more about our AI part of the product.

But, preserve LTL pass grades, back from Canvas two, and that happens, you know, you can do auto auto grading. This is a big deal. At scale when I was talking about three million learners. So a lot of them, you cannot assess it by manually. So you can set up all this auto grading infrastructure, and then they are those get passed or you can set manual grading infrastructure, peer reviews, leader board, leader board is an interesting one.

If you go to our website, you'll see a case study from Cornell. There's a machine learning class. There is a five hundred student or six hundred student or whatever, and they they submit their work, this machine. So they're given a dataset to use it to prepare a model. And then when they submit the work, we will test their, will test the model again, the hidden data set, and depending on the quality of the result they get placed in a leader board.

And these students will spend like submit the assignment thirty thousand times or whatever to just try to move up. I don't know what they get for Cornell, but I remember Stanford was doing this something similar. And the prize for the first praise or they get to have lunch with the professor and the, you know, and I had no idea that it would be that enticing, but integrated with SpeedGrader. That's another one which, customers like our, in a teacher's like a lot, so you can see the student submission right there. Run it in a SpeedGrader, and you don't have to go outside Canvas.

So here's the, you know, the the core capabilities when we think about our labs, delivered the, you know, bright level of abstraction. And I wrote that bullet and whatever, and I'm kind of forgetting what I meant by that. So We'll just skip past it. Budget management. So that's a crucial thing, whatever, right? Because some of these resources cost money, like GPU or cloud computing or Open AI credit.

And your options, sometimes can have really bad options. You can ask a student to go create an account, but, Sometimes they don't have credit card, especially if you're talking about edx, you know, running into all different countries internationally, Sometimes they have credit card, but they'll forget to, you know, they'll they'll forget to, like, shut their machine down suddenly, like, the two thousand dollar bill showing up on the credit card. And of course, if you provide, the access to the resources. You have to worry about the same thing. The student leaving it and you don't.

So every university I've talked to has a cloud nightmare story. Okay. You know, So we deal with all that. In some cases, these these are what I'm talking about an intense, kind of accidental But there are a lot of, you know, classes, which we support in, for consumers where, especially Amazon, where they are given free access or or they will what their students would do is, like, they get two weeks to turn, you know, sort of get a refund. So they will go sign up I'm talking about consumer, not not an academic environment.

They will go sign up. They'll start a crypto jacking job. And, you know, get a refund and, you know, keep getting hundreds of pounds. So lots of once you start giving paid resources, so there's a lot of complexity around here on how do you manage the budgets, right? A rich suite of assessment option already talked about that. Integration with learning environment, that's LTI.

Persistent work area, that's, again, another big deal that you want to. You know, when this learner comes back, you don't want them to have to start all over again. Right, ability to deliver large data set. So sometimes, especially when you're talking about large language models, the data a data science, you might have a tens of gigabytes worth of data, here, deliver it. And learn analytics.

So this is, again, bigger and bigger deal, you know, how do I know how the learners are doing? So we'll, you know, we'll deliver lots of, data. So, here are some of the assessment options that are talked about at auto grading manual grading. So when we talk about manual grading, you know, in a classroom environment, you have to be able to say, lots of you know, we have to be able to assign maybe graders and team projects and peer reviews. So there's lots of capability there, built in plagiarism detection. So Obviously, anything to, you know, you have to worry about that.

Right? So this is not built in, so you don't have to go outside the system, inline comment peer reviews, leader board talked about all these assessment options. So there are about three million learners, platform. The last one I don't know how old this data is, but, cloud is one thing which you do at a scale. You know, hundreds of thousands of labs are being started. We are the go to, lab provider.

Once again, you might not realize it, but if you go to your IDemy and if you're learning cloud computing or going to lots of AWS programs. You're actually using our labs. If you go to edx, you're taking courses, coursera, simply learn in full stack academy. Those are interesting, but those are boot camps. So boot camps has become a newer use case for us once again, you know, there are quite a few people trying to get you from, you know, boot camp to employment.

So we're seeing a lot of that. So, Vocaadium is deployed in self paced, certificates. I think it So, talked about their degree programs, which is, of course, I'm assuming most of you were interested in ILT bootcamps. So bootcamps are interesting. Like, you know, it'll happen like forty or fifty students will show up and there's an instructor and they have ninety minutes of time.

They are teaching an instructor saying, okay, go ahead and start your cyber ranges, and they are you can't have a problem then, you know, because one student has a problem at that point or brings the whole class down over, right? So you have to spend a lot of energy making sure that the environment work. High stakes exam during COVID, we're doing a little more and more of that, right? So high stakes exam being conducted in Vocarium. So lots of capabilities built around that. Rest API. So while we do LTI integration, but not every tool has a LTI, So in that case, the integration happens through our APIs.

And also it's used for content management. So once again, the content. If you're delivering content, you might manage your content in a different, content management system. Happen very often these in the git repo and then you will update it. We'll establish the CICD pipeline, so update the Vocarium using our API.

So that's a very, very common use case. So let's talk about the AI part of what, so I think I went through a little bit when we say AI lapse, what do we mean by that? So we mean three different kind of what you're teaching. You're teaching prompt engineering. And, so So once again, that might or might not require any coding. You might just be teaching someone to go send a prompt over to such a GPT or Google or your own model.

Fine tune a model. So, you start with a, you start with a model and this is essentially happens in one or two ways. Which is you are using as an API to fine tune a model, so which is open AI, the chart GPT. So they provide you, you can download their model. But you can use that API to send data and you can fine tune it and then you can access that.

Then or, if you're using, for example, Facebook went out there and open source their, llama two, the the their foundational model That one you can download it because in open source, and then you can, you can fine tune that. So and then find your model using API, I guess, just talked about it. Right? You can and then, one thing which which is not mentioning that you can, you can create a model from scratch, right? So you can go out there and train a model from scratch. So those are sort of the laughs that will be, you know, will be showing a little bit. What problems are we trying to solve, with with our labs? So first of all, from a learner perspective, they no need to create their own open AI account, or they have to have provision GPU resources.

GPU resources are becoming as you guys probably have been following it like becoming a scarce resource, expensive resource have made a Nvidia trillion dollar company. So one of our core, sort of effort right now is to how do we deliver the cheapest GPU resource possible for learning. So we'll go as bit. You can configure all these things. We will bid on a spot market and you can configure saying, okay, you can okay, I'm willing to wait for a couple of hours for my job to be done, but if I get, like, a spot price, and spot pricing can be, like, seventy percent cheaper.

Right? So but you don't want to and you they can do it with a single click, and I'll show you in a demo. So the learners can just go out and say, okay, run this training job on a GPU. They don't have to learn this complex thing. How do I launch a GPU server? How do I move data? How do I get it back? For instructions perspective, You can, we talked about budget. You can use, you can figure out, you know, mostly use the most cost effective GPUs or large you want, free eye.

And, And the environment, the last part is actually a big deal. Someone was mentioning this to me yesterday. In this set, it won't show up. That, you know, they have to create like for their for the university, they have to create thirty five, thirty six different environments or whatever. Right? Yeah.

I guess you you mentioned that. Right? So so we we went through this same learning and we said, okay, this is getting pretty painful. So we need to go and clean that up. Right? So how how we make it easy to deploy these, customized environment. So this is, the presentation.

I'll just going to go through. Any questions before I dive into a little bit of a demo? Yeah. Sorry. I'm not a technical person, so forgive me if this is an odd question, but if, look, the locarian environment is, integrated into campus. So for students, work and data, cross courses, can they carry their work across courses, or do they have Yeah.

I mean, that's a good question. And unfortunately, I mean, the answer is not only a single course. Actually what we, a single project or single assignment that is completely isolated the environment, and that stays there. And, so it's completely independent. Now going back to what are the things which we are implementing, you know, and this is something, you know, teachers like go back and four, they're like, part of it is, this is gonna be painful if the internet is going to mess up my So, But we are, in our latest version, we do have a file system, a global file system.

So the learners can say, okay, you know, here is my data for this I just want to copy it over. I want to save it from my personal space for whatever reason. So they can do that. But by default, every assignment is comp data resource is completely isolated because, we sort of need to know when to launch it and when to clean it up. Okay, the internet, is there a different, I mean, I just connected to the structure gone.

Maybe I'll try mine. Let me do this. Let me try my cell phone. Let's see if it works. Heart part here.

I was like so excited that when I could, when I could use my laptop, I was like, oh, people show it? Like, I don't have to just ask everyone to visualize how it Oh, come on. Okay. This is this is going to render. Yes. Yes.

Indeed. Yeah. I can't even say, hey, stop by the booth because, you know, unfortunately, the booth is yeah. Go ahead. I didn't ask another question.

Yeah. Thank you. And then you please all of you. Have the ability to set funds for a course. Is there also an an admin, role there? How how does your budget work? Questioning controls? Yeah.

So the so there are two levels of control, as you know, in fact, three level control. Vocaariam goes out there and says, okay, this organization, we have business deal. They can use GPUs. And the admin says, okay, this particular course enable GPUs for that course. And give the instructor ability to, set budgets on art, right? And then so that's, so that's kind of three levels of control that we passed on.

Seen that, funding for for us for AWS. So we have an on prem Jupyter hub, so everything runs local. We don't have to worry about for that. We also provide AWS resources -- Right. -- but the credits are sort of in flux in terms of their their process for giving out credits.

Correct. Correct. And they sort of wanna hand that up to the individual instructors to manage all of that on their own. Is that what's happening here are instructors going to these cloud providers, getting funds, and then put them in themselves, or is it all coming down the top So, so there's, you know, in vocational, you go different labs, right? So, so when we're talking about our JU AI lab, that is running on actually Vocation infrastructure. And the reason being, because you know, as I was showing earlier on that we can take the same file system for interactive and it can just be mounted on GPU.

So the way we build you for that is, you can we were you can so you say UCSG can define their own cluster, and you control things like, okay, as I think, spot pricing, this type or size. And we will when we launch a UCS cluster, we'll tag the resources, and then we'll say, okay, UCSD will launch, you know, here is when we launched that. So that's the. Now going back to the cloud compute, so the cloud computing case, the in fact, it's interesting to ask the question. Today, what happens once again happens at the organization level, we, we get a a payer account and then we generate like, you know, up to five thousand member accounts.

And those member accounts are stored out for every student for every every part. But obviously, you can run out of the you know, you can run out of accounts. So what you do is like there are kinds of policy. You can set this account reuse policy. You can say, you know, for three days, if, if, And the reason why three days, because if you have to accrue budgets, sometimes Amazon, you know, the bill comes a little late.

So you can say, Hey, just keep this around for three days and then you can reuse that. Right. So, so that's kind of how we are managing. Generally, it's it's happening at the organizational level, the, the budgeting, right? I mean, the, the, the, the resource. Now, I have heard this, can the instructor bring their own account? I mean, obviously we can set it up.

It just that hadn't gotten to it. So very soon I will, I will start like talking about the demo because I just have to give up it, and it but, but let me, in the meantime, any other, let me, I'll try the instruction. Again, there, so even this didn't work. So any, any other questions? So anyway, so if the, no, if the demo had worked, So what would I have shown is like as an iframe inside Canvas, this Vocaadium notebook. You can launch it.

And how many of you are familiar with Jupyter notebook? Okay. So, so it looks like a Jupyter notebook, but one is, one little difference. We have added something called a prompt cell type. So just for if someone is like, you know, so you can create a prom cell type and you can configure it by there's a default where does the prompt get sent to right now is GPT three point five or something, right? And you can go and say every learner gets, two dollars worth of open AI credit. And then, so they write a prompt and when you run it.

So this one, you can do it. We don't have to be technical at all, right? You don't have to know any coding or anything. And that's the reason why we created a prompt cell type. You can write a prompt and you and you say run and, and it will send prompt to your whatever you have configured it to your to your model and get the response and the output cell, it would show that. Now it can be submitted.

Just like any of our notebooks. And as an teacher, you can see the response to it, and you can you can give comments saying, okay, you know, your prompt could have done this out over right? So that's kind of one thing which I was gonna show. The the other thing was that you can So let's say if you're working on a Jupyter notebook and at some point and that's running on our interactive Docker containers as you know, and you can done the debugging. And that resource is, by the way, including in our, so all of our interactive resource, we don't bill for it separately. That's kind of part of try to optimize like crazy ourselves by launching a spot, this side or whatever, right, to make sure that and a shared server.

But then you say run training, there's a button which you have. Which takes, takes a notebook and, and if needed, it will launch a, GPU and, depending on what you have set up, it will try to find spot. It can cue it and make it If you say it will just wait, for maybe you can say, Hey, for a couple of hours, I'm okay, you know, waiting for a couple of hours if you have to find a spot because spot as I was telling you that I was actually surprised that the last few months, if you look at Amazon, it might change. It's like you always get seventy percent savings. So there is not that much difference either you get it or you don't get and, and then we run it.

You just say run training, and then the output, you know, gets dropped in writing your same environment. And then once again, you can configure your prompt to go, to direct it to your own model, right? So that, and that was what we are releasing. One thing which you are not releasing is just coming down in the next release. So we do a monthly release is to be able to share say that I want to share this, this particular cell, prompt cell. So we'll give you URL.

You can share it with anyone outside of whatever and then, you know, they can run the the model and try it out and all that. Right? So there are other ways to do this. I mean, so if you got the familiar with hugging face, they they do it with the app called Radio and all that, but that, you know, it first likes too much work. This one you just say, okay, here's the cell. Just share the cell for me.

Right? Yeah. So, you know, that's was my demo. But, the rest of it is of course, you know, integrated to speed grader. You can do auto grading and you can set it up hopefully fairly simply just by say, Here's a assignment, and what's the working locarian technology, what's the lab type, and it says there's a notebook, it's a Jupiter, it's a VS code or whatever, and then you say what additional resources I want to provide to this learner, and you can pick AWS, Azure GCP, data breaks, and then we'll say how much budget are open AI? And he says, okay, how much budget do you want to give? And, you know, the, and the learners see the budget. It dynamically updates It was a wonderful demo.

No, unfortunately, but that's the problem. This is the boot just got done at nine o'clock. So I was like, my entire presentation was based on, hey, you can stop by the boots, you know, I can show it to you. Now, you'll have to say, you have to ask we are virtually, like, you know, obviously we'll set up, you know, set up demos and, you know, yeah, for a call and thing. Yes.

Are you are you serving as a platform of a utility or do you have a curriculum as well as utility problems? Yeah. So, you know, until right now, there was, it was a pure platform play. So we brought it up. It has to be completely configured by the by someone. Right? So there's an authoring interface.

Although just because we are seeing so much, you know, and that we I've always edited, I've been hesitant to create content like so much, you know, so many different things are being taught over, right? Like, one of the case studies you will see in, and I was surprised to see, you know, Steve Toronto professor the teaching quantum computing, right? And I'm like, I don't even know what that means, but somehow it being our use or if you delve University is one of our big clients, like seventy or eighty courses they launch, you know, and occasionally I go and see the chemical engineering, quantum mechanics, you know. So that's why, but going back to so, but we have decided to start creating some foundational lab for the basic Python data science machine learning. So we are beginning to do that. Now how will we deliver it? We are not even sure at this point, but auto graded labs and and because one of the things which are beginning to hear from a customer is like not for education perspective, but for assess skill assessment perspective. Sometimes the corporation, sometimes actually university is saying, Hey, my incoming, my incoming student, it wouldn't it be great if I can just, quickly like, assess their skills so that I can, you know, help them find the rights classes or, or maybe even have a little boot camp to say, Hey, you know, you need to just just do something, you know, pandas.

You cannot really recon pandas. Let's do it two days of, you know, pandas or whatever. Right? So anyway, so that so it's, but courseradex, everything is the same platforms. You know, the way we think about it, is kind of hard to say that being it, like, instructor conference, but we are in the tech stack that you need for education, we are solving. We like to think one of the hardest problems.

So for, and I am saying, like, obviously, the instructor guys would probably disagree, but, we are solving on the hardest problem. Because you have to be able to deliver videos, you have to deliver multiple choice, you have to be able, but, the part of the tech stock, you have to be able to deliver these labs, right? And hopefully as seamlessly as as if you're asking the student to write some text, you know, or essays or whatever. Right? So that was our first mission. We feel like we have done a decent job. Now, we're trying to do some content here, but don't consider, we don't consider ourselves as a content experts at all.

Yeah. So no partnership with any, textbook publisher at this time, or is there anything like that plan? Yeah. I know we have conversation. I'm just trying to think. Do we have partnership with the textbook? No, we don't right now.

But, you know, it's like, you know, we have conversations all the time with the usual, you know, the Right. Now, unfortunately, we we don't have, you know, we don't have any our key partnership, we really spend energy on the learning platform providers like edx course that are simply learned, Ameritas, you know, I'm forgetting a few, you know. So, so we went through that was a big focus because they were know, so we have a saying about three million learners. My goal is to get to ten million. And I'm like, okay, where do I find learners? And that, you know, that's where the, that's where the you know, the high volume.

And of course, but what has been really fun as a singer at AWS Academy. We have hundreds of thousands of learners coming, actually they are there as a part of the degree effort, right, either in, teaching universities or community colleges and Arvans and so all over the place. So that has been pretty exciting. We're very proud of that. Any other question? Well, so in terms of, follow-up, vaquarium dot com, there is hopefully you can if you But I think we we might get their eyes.

We do you think they're gonna get their Ophelia and their sheep, in our marketing team? So I think we get their contact information. Because I saw someone actually in a sort of scanning game. So, yeah, so, hopefully, we can, we can, we can also reach out to you, but you, you know, if you are interested in more demos, more information, since I was not able to do that, please go to our website and ask for a demo. And, otherwise, thank you so much. It's been great.