Large Language Models Bootcamp- Information Session

Key Takeaways

e yes Raja we're live now okay thank you SOS we will go ahead and get started just I will just set up a few things here on my screen okay um so we'll go ahead and get started uh welcome everyone to our information session for the large language models boot camp um my name is rajal I'm the chief data scientist at data science Tojo and I'm also one of the read instructors at uh the llm boot camp and also some of the online offerings that I will be mentioning uh in a moment um so we have done this before perhaps you are you have been here before uh in this information session um so I will go over very quickly uh about the logistics what the boot camp is about and what it is not about and and some of the online offerings that we have introduced and we have launched the registrations have opened um so we'll talk about it uh so we are one we are one of the oldest players and one of the probably most notable players in this space um in terms of data science and machine learning analytics uh trainings uh across the board so we uh we do training all the way to very Hands-On U hardcore machine learning uh AI business analyst data analyst type uh professionals uh and then also for product managers marketing managers uh you know the general ecosystem that actually consumes uh the data products or they are part of the a datadriven organization um Enterprise trainings we have a sizable uh part of our uh our business is actually Enterprise training so if you're one of those uh individuals who may have a team that wants to be upscaled uh please reach out to us um and this is our numbers more than 11,000 graduates and this is I'm talking about face Toof face right so we have taught people in a face- tof face interactive setting uh this is not self-based courses I mean self-based numbers I don't know what those numbers are those numbers are going to be very huge um so uh overall why this boot camp so um uh last year around this time uh around February March time frame um we started doing some Services work we started doing some Consulting work for some companies um that wanted to do build large language model applications well um we have been doing a lot of machine learning work for companies but this was the first time when we started doing some of the llm application work so we started uh and in some cases some companies had already built these systems and uh they were scratching their heads now right so what do we do uh so we have started it but there are some challenges uh the go the bars of the world the chat gpts and clouds of the world they make it look very easy right so generating this content this social media post that uh poem that question and answer that I wanted to have or wanted to get Maybe do my homework get some help in writing here and there that seems very easy but when you try to deploy these applications in an Enterprise there are some practical considerations there are some practical limitations and what are those practical limitations uh so it may uh chat GPT May uh seem to be free but there is a substantial cost that openi actually incurs in hosting those uh and then allowing you to use chat GPT for free and when Enterprises actually start um start to um adopt um these these tools the large language model tools they have this uh realization it is tough it is tough to manage uh you know manage all of those things uh you know uh starting from uh the cost you start with token usage and cost context window limits right so uh even if you have a 200k context window model you still have some limitations there right so because your real world data is not going to actually be it is not going to obey your business constraints what do you do in that case there are regulatory challenges there's you have proprietary data uh you know regulatory challenges I mean I can give you an example right so I'm not sure how many of you are familiar with this example uh Air Canada was they had to face a customer in code because uh uh one of their chat Bots gave uh a bement fair to one of the customers and the customer uh purchased the fair uh believing the chatbot and when the customer went back to ask for that discount Air Canada said what discount um and then the matter went to court and Court ruled in favor of uh the customer because they thought that Air Canada is responsible for their large language models spot which is uh I mean if you look at this we are living in this new era of uh uh large language models and generative AI where you know our the nature of our challenges the business challenges is going to be different then you're talking about uh you know any proprietary data so any proprietary data leaking uh or getting in the wrong hands um now uh when you asking questions uh are are you getting the questions in time or are you getting those or rather are you getting your responses in in a timely manner or not uh what about your data and AI governance um do you use open source models or do you use uh close Source models um there is this challenge of lack of reproducibility so can you reproduce the results or not uh hallucinations is a problem your your knowledge base it's constantly changing uh how do you evaluate a model so I can keep going we realize that it is non-trivial enough that enterprises would actually need to be upskilled and uh we decided to launch this boot camp and the boot camp is actually uh the only boot camp in Industry at the moment we are the the first in the world uh and the only boot camp in the world that teaches you end to endend yes I mean there are courses uh on um on observability here and maybe embeddings here and Vector databases here there so these courses do exist but we are a comprehensive course that covers everything end to end so you start with the very foundation and end with really the a complete application you build that application within this boot camp and that is what our differentiator is we have been around for a long time close to 7,000 graduates from our close to 7,000 gr graduates from our data science boot camp and we have a stellar reputation when it comes to data science boot camp and now we have launched the similar product a 40-hour immersive inperson uh program and with a with the Capstone project on the last day and uh well everyone cannot travel for an inperson boot camp so we were asked about what about the online courses um and then we have launched we are slowly rolling them out we have announced two courses I will show you what courses we have announced uh and this is going to be the same Interactive Learning um they can the courses are uh 4 to eight hours some courses are 4 hours some some are eight hours shorter duration uh of course the cost is uh somewhat more affordable because you don't have to travel and then we don't have an inperson commitment involved uh and then we have tons of free self-based courses I mean come join the learning platform just interact with courses and they are there for you for free so I will first go over our um our the boot camp that we are uh we have so in terms of uh the curriculum um the the curriculum has been designed by practitioners I spend almost almost 50% or maybe more uh than 50% of my time on um on a product that we are building and uh my daily meetings involve uh you know uh talking about you know how to optimize our hybrid search for you know the rag application that we're building right so you know you know how do we put proper guard rails here right so talking to customers so the curriculum has been designed I've been involved in designing the curriculum among other data scientists so one of the key differentiator is this is not actually some course that is hey I I read about I read that paper I don't know how it works but I can tell you how it works right so and I've never done it right but in our case anyone who teaches this training or this uh in in this trainings either uh if they are from data science Tojo they are practitioners and if we have a large number of people um you know I think we have a instructor uh combined instructors and speakers we have about I would say 10 10 people teaching in in on different uh sessions and they are coming from leading companies in the space we have a speaker from mongodb we have a speaker from um vcta we have a speaker uh uh from U ylabs which is an observability um the company that does uh create makes this Tool uh for observability we have people from uh stripe and I can keep going I think I have I have a slide where you will see all the instructors um the other thing is um when you sign up yeah our learning platform it has all the tools that you need as long as you have a computer with a web browser we what we have done is we have prepared everything for you uh and then you do not need to worry about open AI keys or you have don't have to worry about you know licenses and dependencies um on for different packages uh you just click on the lab and we have hundreds of those in this uh in this uh in our course uh that you will see I will show you all of that I will walk you through everything and then uh you know there's a lot of Hands-On exercises and then the the focus is I would say 50/50 we talk about Theory we discuss engage uh uh and then once uh the context is set up we go to uh Hands-On part so pretty much everything if it's not going to be only hey this is embedding right so this is Vector representation this is the tension mechanism Transformers you will actually have a lab where you will do it when you talk about Vector databases you will have a lab for Vector databases and so on so I will explain everything in a moment um and then as I said just bring your laptop we are including when you register uh up to 500 US dollar of credit for all software and cloud services uh so we when you do a fine-tuning exercise uh you will be given a GPU Cloud um to actually u u to actually um fine tune your model uh if you are uh you know you're setting up semantic cach uh you know in mongodb we will actually uh you know give you the right resources so basically you do not have to worry about hey do I have the software what software do I need to install and all of that everything is taken care of our learning platform is amazing we have installed everything uh so everything is there uh included U and then uh we have guest talks uh so it is not just uh data science JoJo um explaining and talking about things we have guest talks from industry uh speakers in industry who have actually built real solutions and they have seen Solutions succeed and they have seen Solutions fail within an organization for various reasons uh the solutions could fail because a solution can fail because of you know many reasons the customers don't like it your Workforce does not like it so all those things they are actually essential in being a good technologist uh if you do not understand the business side of it the cultural side of it you know all the other uh non-technical issues um it does not really you're not a great engineer unless you understand all of that so we have a good combination of all of the things it's a very balanced curriculum we don't focus on coding coding coding we also talk about you know other issues that you may not have thought through um okay so let me see I'm going to go over the at a very high level I'm going to go over the this architecture this is the reference architecture by no means this is complete but uh this is just to give you an idea that we actually use uh we actually will be teaching you everything across the board um maybe let me take it back I should not say everything but most of the most of the things that exist um in this ecosystem you will leave with a very good decent understanding of what is happening so starting with llm apis both from um either you're using the llm apis or you're using uh an open- Source model that you're deploying we have both happening um um we will be using open AI a lot you will get the keys and uh you'll be getting the keys uh and accesses to the services when you're using open AI when you when we um when we use l 2 uh model um we have a fine-tuning exercise for L 2 7 billion parameter model when we when you do that um you know that's open source side will give you the infrastructure and everything uh and then in addition to you know the the setup uh and the uh the instruction material um so the in if you look at the architecture so the llm or the foundation models they are a key component in this ecosystem then uh Foundation models alone are not enough you need something to create those embeddings right so and we will talk about embeddings uh you know we'll talk about embeddings in detail then then we are going to actually uh also talk about how do you store these embeddings and they there are going to be Hands-On exercises you create the embeddings uh using whatever models uh whatever embedding model you're using and then you we'll also talk about storing and indexing optimizing the index of uh in a vector database and then how do you retrieve how do you build a sematic search engine so uh if you look at this this is a typical architecture for llm application and we uh I'm I'm showing you I'm sorry I flipped the slide too early so we are going to actually talk about this then uh uh you know there is this idea of a semantic cache there is it here there's this idea of a semantic cache and semantic cache is well uh every time when you interact with the large language model you uh you encounter some cost of uh token consumption right so gpus are expensive uh compute is expensive so semantic cache is that layer where you sometimes you you cache your results based on semantic similarity so we'll talk about first of course by the time we get to semantic cache you will have you know it's halfway through the boot camp by the time you get there you will actually know why do you need uh to have what is a sematic batch first of all and why do you need to have a semantic cache it is basically safe cost and reduced latencies in your application uh then uh there is a whole area of how do you actually collect your logs uh how do you um how do you actually uh uh collect your logs put guard rails around your um around your application so it does not um you it does not respond to any prompts that are toxic they are inappropriate or you know they don't meet your Enterprise guidelines or your model does not respond to any queries that should not be catered um and then we have a very detailed uh we have a very detailed session on Lang chain if you look at this the entire ecosystem I'm sorry I'm keep keep switching back and forth because the this really the the entire ecosystem is quite complex uh you know uh yes I mean one can go and attend a course here a course there a 2hour course here and you know a 1 hour course there but what we are doing here is we are doing this comprehensive end to endend uh description end to end understanding of things and putting the connecting the dots together right so for instance right so I can go and attend a course in Vector databases but I have no idea how to use them I mean where where do they set in my application so what what we are doing here is we are not only teaching each of the modules here we are also teaching you how do you stitch them together to build a working application how do you put once your have applic application is there how do you connect it to different data sources you know so we uh or how do you connect it to different types of models and uh how do you optimize how do you how do you put guardrails on it how do you make your application secure so think of it as a comprehensive endtoend uh discussion or end to endend hands-on experience where you will learn everything across the board so uh I think in instead of going from here so this is the technology stack and then I will switch to the our learning platform so this is the technology stack that we are using at the moment we have few more Partners who are coming in but this is a technology stack that we are using uh you know uh we when you build the app the app is deployed in streamlet so you will learn a little bit of you know uh app deployment uh then um you know when you do your uh fine tuning of your uh model then you will be using runp pod they are they are a partner for the boot camp so the thanks to them so we are giving you will be giving you uh uh GPU Cloud credit for uh for that exercise uh vcta as a partner vcta is uh you know they have an application for some of you who may not understand but I mean this is your rag in a box uh which is retrieval retrieval augmented generation in a box uh then mongodb is partner they are I think most if you are in Industry you know what what mongodb is but mongodb also introduced a vector database functionality and now we U we will we have a session with mongodb on Vector databases and how to use Vector mongodb as a semantic cache symol AI they are a great partner for us uh and then um there is this uh this concept of domain specific models so uh symbol has this model which is nebula so once you've understood the idea of embeddings vector databases and fine-tuning now how do you actually build this domain specific uh um domain specific uh uh models so this this is also I mean personally I this is one of my favorite talks uh because I mean car who teaches it and is a great great instructors instructor uh so what are we going to do here we are going to actually set the foundation um we have once again in this area we have L Sano for those of you who know him he is our uh guest speaker for that he's one of the best people if you want to learn Transformers and attention mechanism so he teaches uh Transformers and attention mechanism and what what really is all of this about what is an llm and you know uh where why do we need a vector database and why do you need uh uh embeddings and how does attention mechanism work I mean what is an encoder what is a decoder and all of that so uh we talk about that uh then we go and talk about uh you know embeddings how do you create embeddings and I will show you the Hands-On exercises in just a moment and how do you uh how do you actually uh store those embeddings in Vector database how do you optimize uh the in how do you optimize your uh the index where the embeddings have been stored uh then uh building a semantic search engine with multi- lingual uh search ke capability prompt engineering we spend uh time on that uh uh L chain we spend quite a bit of time on Lang chin I think I should probably probably switch here so this is the learning platform uh that you get access to when you sign up for the boot camp uh and you can see on the left um you have uh you know we have some prerequisites prerequisite in the sense some refresher uh in Python and all of that that's all you need if you can program in Python you can attend this boot camp so I know that this is a common question that people ask what are the prerequisites we have had people who even did not know were not very good at programming uh and the way it works is um they were able to actually um they were fine attending the boot camp they did you know they were okay and the way this works is if you look at this when I click here um uh the way we have structured this is that you have uh you know uh anytime we discussed a topic uh within the topic we have these links to these labs and when I click on this lab I'm sorry I clicked on the long wrong lab here let me launch it again when I click on this lab you know I as long as I can you know I can I'm running this so many people who did not I mean people who were technical product managers they were not necessarily the best coders but they wanted to attend the boot camp just to understand just to guide the product they were actually very happy and you can look uh look at the testimonials on our website real people uh by the way I mean there's not just some uh we don't have any stop photo models or some John do and Jane do giving us reviews these are real people with real LinkedIn profiles go and talk to them uh and they they are our Champions so if you look at this I'm I'm I can run run run and I can run this and then if you are someone who is uh if you are someone who is more um technology or more programming Savvy you can go and modify uh you can go and modify the code and if you're not uh that's okay because uh you can still uh you can still um run this code and understand at least at a very high level how does this whole thing actually work so if you look at this this is our lab for prompt engineering just over basic first lab in prompt engineering so I just put the API key for my my openi API key in this this part of the code and then if you look at this so this is all included in so you will get the open API key in class when we are working you have access to this there's no setup required on your site and if you notice this is my jupyter notebook this is a persistent storage here this this is a processing storage that you you can see here and and now if you look at this here uh you know uh so everyone gets their own dedicated personalized storage you don't have to worry about you know where will I install this do I know which packages we have taken care of all of that for you and then you are you are running running running and then over here uh you can see that as I run the code um you know um this uh you can see I just this is just an example of generating an image uh programmatically of course examples are going to be somewhat more more involved than this um and then um and we are going to go into a lot of these exercises maybe I will show one more um one more uh module here that is the Lang chain module I think I had it open somewhere here yeah this is our Lang chain module for instance if you look at this we talk about pretty much all of these um uh you know we talk about um different how do you connect uh your application different types of models how do you uh how do you connect it different types of data sources different kinds of parsers different kind of extractors uh different kind of chains Let me Give an example right so there is this example of uh simple sequential chains and you can now see even if you're not a Corder you know how to click on this uh we have talked about what is the purpose of chains how to use them and now if I run run run the same drill sorry I think I should have given the key here uh I will copy and paste and and run here and then if I run it you can see that I'm I'm I'm using a gbt 3.5 turbo um and then this is an example I'm not sure maybe most of you may perhaps don't know what chains are but uh you know this is basically uh chain is connecting uh instead of giving a single query you're basically connecting two queries uh and or the two prompts or two uh inference queries that are connected with each other and the second query depends upon the output from the first query in this case I think this is uh um this is and I'm running this chain and this is you can see that you you see that how it is basically U running the chain and giving me an answer so uh one of the most common questions that we hear is can I do it right so it's uh I don't have a background background in machine learning uh um or I I'm not a greatest programmer I'm just a product guy um or product person uh that is okay uh I mean if you if you can read English python is you know if you run run run so the as long as you're coming uh with an understanding what do you want to build uh if you're coming from a generally I mean you you understand uh the generally I mean what you want what application you want to build we have had people who did not have a very deep coding background they were okay if they in many and you don't need to have a machine learning background we actually do discuss uh some of the machine learning uh not in depth but enough Concepts that are needed to actually for you to be able to uh uh to be understand uh how how things should be done and uh um so we uh we talk about a lot of other things uh I can uh you can see that there is fine-tuning is there uh we talk about uh uh this uh nebula the the conversational AI uh we talk about lank chain we talk about uh observability and evaluation how do you evaluate a model uh because in in a fraud prediction model in classic machine learning and fraud prediction or customer turn prediction your answers are clear either it is a fraud or not a fraud in this case two answers or two responses can be similar right so some kind of uh you know it could be uh to responses could be you know a human would say yeah these are same responses but the the words that have been used they are not the same so how do you evaluate the response when that is the case so we talk about that we talk about a lot of deployment related uh uh related uh issues and we spend a lot of time in discussing the nuances uh mostly we even though we spend we spend about 4ish hours on finetuning but the rest of the rest of the booth Bo camp we talk about um uh we talk about all the all the nuances and the challenges and really optimizations in a rag application right so how do you choose the right chunk size how do you augment your chunks you know different kind of uh uh chunk uh chunking strategies we talk about that right so we talk about hybrid search ranking Fusion I can keep going I can keep throwing terms here but the general idea is that uh it is um uh we we actually go in quite a lot of depth uh but we build slowly right so it's it's uh it's it starts with the fundamentals and we slowly build the momentum right so we don't go to Vector databases unless until we have understood embeddings and we don't go to semantic search until we have understood embeddings and Vector databases we don't go to um lank chain uh un until we have figured some of the uh foundations out and uh we don't go to observability and guard rails until we understand in general how an application is built so in general it is uh I have to say that I mean very well thought out I mean because we do this uh for a living on a day-to-day uh basis as I said I spend 50% of my time in uh in actually in on a product that we building so we have an actually a product um that has customers uh some um some big companies that are using our products so we know how these products are built and we know how uh what challenges companies run into so if you are uh if you are uh aspiring to build uh llm application at your company um this is the boot camp and this is very practical very Hands-On uh let me show you a few more things I know there are a few questions but before I go to those those questions I am going to actually I'm going to go and show a few more things so uh one form of exercises that you saw are um uh this year when you're basically you know uh clicking on this and the Jupiter notebook pops up like this um we have actually been building a lot of other things I mean so let's say this is our prompt engineering exercise so you know and now what we have done here is uh we have built uh these tools where you are able to uh you're able to um you're able to practice prompt engineering right within uh you know without even if you don't know if you don't want to write python code you want to practice just prompt engineering without the coding part of it uh you you can just uh go and write it down so we have these sandboxes not just for Python programming all but we have sandboxes also also for our also for prompt engineering so let me uh demo it here right sorry uh write uh High to about geni right so so now you know and we have a lot of exercise I mean definitely that the exercises are going to be uh something more than um more than what I'm talking about and then now you you can you can see uh it is it has come up with high cool and then if I uh expand this you can see I can actually choose the number of tokens I can choose a temperature let me you know you can play around with temperature and then it is going to um you know just give me a bunch of different answers uh based on this uh different settings so we have uh we have the curriculum a very carefully designed curriculum the instructors topnotch instructors uh teaching the uh the boot camp um then the right Tools in place uh so you don't have to worry about uh you know installing things uh so it's all set up I mean so you have you just come in and start learning let me go back and see uh what else did we have I mean I've talked about most of these things here uh Enterprise created applications and then I don't know if I mentioned but there is an actual application that you will build build uh on the last day of the boot camp so uh the the the five day boot camp that we have on the last day of the boot camp you go from Monday through Friday you start at 8:00 a.m on Monday and go all the way to 5:00 pm on Friday and then uh on the last day you are going to actually build an a rag based at llm application um we will give you the code we will give you the infrastructure you using our Cloud subscriptions our accounts uh except that the GitHub account is going to be yours of course so and you will actually leave with a working application deployed in streamlet cloud and we will be giving you everything and then it may sound um it may sound like a lot of work but even if you are not a programmer streamlet is actually uh something that uh that Shields you from that uh you know coding uh and then even if you're not a programmer you should be able to actually have a running application right so the the way we have we we have been doing this 11,000 people is a lot of people that we have trained and after that many people that we have trained our infrastructure is very very mature and our processes are very mature and you should be able to actually even without any coding background you should leave with your running application uh and this is our lineup uh so I'm one of the instructors of course that's why I'm here Lewis actually is going to be uh teaching uh Transformers attention mechanism offer is from vcta so he actually talks about you know this idea of rag in the Box caric uh uh is um he will be talking about um generally cover the conversational AI uh and um you know domain specific models uh P will uh is from mongodb uh you know he talks about you know all the mongodb related features uh you know at this time we'll talk we are introducing this Lang chain and and semantic caching using mongodb so this is a brand new uh brand new new module and this is also one more thing since we are practitioners we are passionate about what we do so it is it is not something hey we have to do it because we get paid for it no I mean we enjoy it right so this module um I on mongodb it was wasn't there in the previous boot camp and now we are introducing it in the new boot camp because I saw LinkedIn post hey I I reached out to mongodb hey guys you want to I mean because I like it and they said yeah I mean we would love to so you know that so the curriculum is constantly being updated uh we living in a very interest interesting time where um you know I wake up and every single day I see a social media post uh on LinkedIn uh where I find out uh especially I follow both Lang chain and llama index I mean great uh places to learn uh and then uh we are going to be like every day I learn something new I mean some of the problems that I working I'm working on internally uh some of the problems that I'm facing and then I would go and um you know yeah I mean this is great so someone is coming up with a solution and then in many cases we incorporate those interesting learnings interesting uh lessons into our uh into our boot camp so the boot camp is constantly the when we started the boot camp about six months ago the boot camp that we started with it is uh the boot camp that is now it is very different and the other thing is uh thanks to the learning platform that we have and if if you attend the boot camp and three months down the road if there is any new talk any new tutorial that has happened you actually uh you know that uh those talks are uploaded uh in the learning platform so you can catch up uh and you have access to the learning platform if there any new Labs any new material any new exercises new coding new python notebooks are added you have access to that uh for one year since uh starting from the first day of the boot camp um then Hamza Hamza is a also you know some of you may know him I mean he's he's running his own uh startup in exactly the same space Sophie uh Sophie is data scientist stripe she talks about some of the cultural uh issues and some of the business challenges in uh in implementation this is one of the most uh favorite uh one of the talks that gets the highest rating great talk uh less technical more about uh you know a more downto Earth real pragmatic talk uh but people love it Sanjay is our um senior data engineer uh he is mainly leads the the uh the module on Lang chain uh Bernice from vabs biabs is a valued partner uh they have been with us since the beginning of the boot camp they talk talk about uh in putting guard rails on your uh putting proper guardrails logging uh and then really U basically surrounding your model with more observability and monitoring and uh Etc and Hamza is co-founder of zml um and he actually talks about U how to deploy and operationalize your uh applications um um and these are these are about partner testimonials I mean so um I mean these are uh well-known people from industry I mean they are these companies they are partnering with us for uh this particular uh this boot camp uh you can see that uh you know we have very happy partners and very happy customers and if you notice uh this uh list I don't think this is updated uh we have had people uh coming from not just uh like within the us we have people coming from as far as AUST Australia right so we have had last boot camp we had someone traveling all the way from um either Sydney or Melbourne uh to Seattle so so we have that you know that reputation and trust that people are willing to actually take uh you know a long flight I mean we have people coming from Middle East we have people coming from Saudi Arabia and you know other other uh other countries in that region so our next in next input person boot camp uh is in Seattle uh from April 22nd to 26th so that is there and in addition to that let me actually bring it up here I will bring it here our inperson boot camp uh uh or so this is the inperson boot camp but we have also launched our online boot camp and our online boot camp is it is coming up or rather not boot camp but short courses there is our first online course is this what we call llm for everyone large language models for everyone I'm very confident that even if you don't know any coding you're not from technology you're a lawyer you are uh uh you're a medical doctor uh you are you know you just you're just curious what is this whole thing about so this course is a very short course uh it's a four to five hour course we will actually teach you about you know this whole idea of what an llm is uh and um you know what are large language models you know what is prompt engineering how does it work what do it dos and what are the don'ts of large language models and then you can see it's uh right now it's actually a fairly very steeply discounted actually uh training because they are launching the online [Music] [Music] [Music] professional body uh certification I mean if your uh your your job requires that or and then you can also some companies they require a certificate from a university and for reimbursement so for reimbursement purposes this is also going to be applicable and then this is a uh this is this is the online uh instructed version of The Lang chain module that we cover at at the inperson boot camp EX exactly the same curriculum same intensity except that it is going to be through Zoom or teams uh and then um you can see the curriculum involves pretty much everything that Lang chain has to offer as I said it is uh eight hours of very intense training that is going to be in uh uh instructor Le live instruction and then in this case once again uh I will be teaching it and Sanjay is going to be also teaching it actually Sanjay is the primary instructor I'm the secondary instructor and he he knows you know sanj is he's a wonderful wonderful engineer um and then uh you can see right so these are the past people who have attended U our our large language models uh training and these people are coming from you know uh you know really big companies I mean so they are not you know some and these are real people I mean you can see I mean just go and look them up these people are actually uh you know they have attended the training and they uh recommend us okay so what I'm going to do here is I'm going to go and start taking questions let me see and feel free to if you are online if you on the zoom call I am looking at the questions but if you are online uh through Linkedin or YouTube anywhere please leave your questions in your comments and then uh someone from the team is going to grab that question and route to me uh yes uh can you share this slide deck and pricing for the boot camp please look at the comments we have if you need the slide deck of course uh you will have to request the slide deck so there is uh some uh uh you can actually uh contact us uh there is uh uh I mean contact us using uh the form that we have shared and then we will be actually sending you the slide de sure I mean we slide share slide deck all the time um okay then um I will come to Betty I know you just posted a question I will get back to you um I will I will get to that question let me address one more question before that I am a Java I'm a Java developer and haven't programmed in Python how can I learn before The Bootcamp learning it before the boot cam I see you have a Python tutorial is it available only after registering to the Boot Camp or can can I have it to decide if I'm available [Music] [Music] [Music] BET's question how can the 5 days in Seattle boot camp and the8 hours online be equivalent I did not claim that I mean if I imply I did not even try to imply this so um the eight so the eight hour online is only for Lang chain so it covers only one day of the boot camp so um so there was a question how can the five days in Seattle boot camp and 8 hours online uh equivalent be equivalent in content so I did not claim that um because uh the online 8 hours course is only Lang chain but the in-person course is much more than Lang chain so I hope that makes sense um does the online option cover exact same stuff in inperson boot camp yes depending upon which module right now in the online version Let me actually bring it up in the online version um if you go to Lang chain so whatever we cover in Lang chain about eight hours it is exactly the same as the Lang chain module that we have and then we spend eight hours in Lang chain on Lang chain in person and then we will be send spending the same amount of time um but we are about to launch uh uh we are about to launch the fine-tuning uh course it will be a separate six-h hour course so basically what we doing is the module that we cover in person we are slowly going to roll them out one by one in online setting right now we have only two of these out of all of these modules we have only two of these modules online but whatever we cover uh if we are if you cover fine-tuning so the fine tuning in person the fine tuning module in person is going to be the same as the fine tuning module on online uh if you're covering observability and guard rails whatever you cover in boot camp it is going to be online so we are basically breaking down this boot camp the we are breaking down the inperson boot camp into a bunch of um online courses um so so it is more affordable you know you can break things down you can see that the price point is very different uh and then also uh you know you don't have to travel so that cost also is not uh there uh okay most universities are giving certificates in one year a few months for AIML can you summarize on how this 5day boot camp achieve that it is like uh this program only covers how to build used tools uh so this is a question by SJ or suj I don't know how to pronounce it right so so uh sege uh our uh we have a long history this we are not by any stretch of imagination we are not a new player um so universities uh you know they have their own structure and how they do things um you know they have to go by um two hours a week or six hours a week and then for entire semester and you know in certifications and all of that right um we have had people who have attended um who have attended the certifications at bigger universities they have attended our boot camp we have I mean check out our website I mean if you go under I will encourage that you go under let me just go back here uh on the website go under reviews success stories alumni testimonials companies and this is not as I as I keep saying these are I don't think any other company in machine learning and data science uh learning space has that many verified reviews including the big names like corera udmi right so you know we are actually one of the most respected and most trusted companies uh so we uh because um universities with all due respect I have a PhD I've been on the Academia side right so there is a difference between uh Reading Writing papers and there is a diff and and versus Building Solutions right so so you have so what we have done here is we teach you the the core of it what is needed so you can be on your own and uh um I but at the end of the day I mean it really depends upon what you're looking for uh if you have the time and the energy to you know spend an year or learning um learning those uh topics uh through an online certification uh through a university of course I mean by all means you're welcome to uh it's a different target market what we want what we do is what we take pride in is um you are going to be in and out of this in 5 days and go check it out reach out to those people hundreds of people actually vouched for us that they started uh as a beginner and then now they are building things right so and that's um um yeah I mean but we are not only covering building and uh using tools uh so we we cover uh so when we when we talk about uh any topic we are going to uh go enough into the details we'll talk about the math part of it we we will talk about the programming side of it we'll talk about the algorithmic side of it but we are not going to spend time on proving theorems and lemas and all of that right that's not uh you know this is um you know uh I mean I I I tell this to my own uh engineers and my own data scientists as well that you have to have a Mind of a data Mind of a scientist but bias for action of an engineer because I mean both sides have their pros and cons um so we all uh we don't just teach programming we teach the theory but we also don't stick stick to only Theory we also teach programming so it is a healthy balance between the two so I hope this answers the question I know it you know I've gotten this question in the past the only way to I mean spend the time uh just uh do some research on our website right so and uh check it out right so I mean not just one two I mean there's hundreds of people I mean uh who have who are on camera they're saying that they are the best training that they have over 10ed right so and we know how to how to teach U machine learning and now large language models uh um okay so Betty your question the inperson boot camp is only in Seattle or will it travel around like uh come to NYC and how much is it right so uh the price point is going to be again this about the same bet um uh but you know NYC we have considered our we we used to pre-co we have done like 10 boot camps in NYC uh so we used to do it in uh we have even this year the LM boot camp we were in DC we were in Austin so we may come to NYC but it is more more about the operations and Logistics that make it very hard because uh this is not a self-based learning course this is actually you know uh you know these people are uh you know like this these are top-notch people from industry getting their time and having them you know come in and teach at NYC it is not I mean you're so because many boot camps I mean they will have anyone who can just uh regurgitate the slides is is okay right so me they can just go and teach right but getting those people who actually have done it their time I mean operationally and financially it becomes uh not feasible to actually get those people to actually teach uh in a different city uh and traveling around right so because this is an in-person boot camp um so online uh yes NYC might happen later this year but I don't have a date because we have done it in NYC NYC tends to be you know a tricky one in terms of getting a menu but we may be coming there but nothing uh in terms of nothing confirmed how much is it it is going to be around the same price point as Seattle we don't differentiate in terms of which city it is so that's uh uh I hope that answers your question very um okay there are more questions any plan of allowing students to join the boot camp virtually I wish I could do this mamy but you know if if you I would love to but really uh the in-person boot camp is an experience right so and uh you know it is uh attending it virtually would not even come close to uh you know because we imagine you you so we serve the breakfast right so before the boot camp people coming in they're already actually discussing I mean in the you know the waterpolo talk and you know having coffee together during breaks but still brainstorming together you know people are actually really really absorbed when they are at this food Camp from 5 days lifelong friendships happen you know people are they stay connected job opportunities so um you know you're eating together and then we have a networking dinner that happens uh somewhere uh during the boot camp uh we go out eat together and still you know talking about so I think yeah uh you will be paying for the inperson boot camp and doing it uh virtually I think it doesn't even come close uh to it uh uh so I would uh and then we don't recommend this because operationally you know if we are adding one more bottleneck to on our side what if something happens you know the audio and all it becomes actually a bit cumbersome so no at this point there's no option of joining I mean we would love to actually you I mean it's revenue for us but we uh we very um you know very we are very protective about the reputation and the quality of the training and then for that you know uh we pay attention to every aspect so online is not going to be possible for the in-person boot camp and joining it remotely okay uh you don't live in the uh is there any financing option for someone living in Canada uh M please reach out uh I am not sure we do have some financing Partners I'm not sure if the cover Canada or not but we should be able to to help you uh what is the minimum requirement uh requirements for the machine spec in terms of memory none because when you uh you can bring in your Chromebook and you'll be fine because when you click on this uh you know when you click on uh you see when you click on uh one of these practical exercises in reality this is not running on your your computer it this these labs are running on our Cloud infrastructure so we are we have preconfigured all the packages all the machines so even if you you can run this on your on technically I have I've run this on my phone as well right you can run it on your iPad so there's no machine requirement and this is exactly in the past what has happened is people would come in uh and then uh hey I can it would not allow me to install this package what should I do and then you are basically um in a different city we used to carry extra laptops because we knew someone is going to get stuck in this uh and or sometimes you're spending an insane amount of time just resolving their dependencies of packages conflicts and all of that if you notice this mamy here uh the you are going to get a dedicated space in our boot camp environment um in our um um environment and then any code changes that you do they are persistent this is yours to keep you know and then you can can actually create as long as you don't start mining bitcoins here right so I think uh you are uh you know you're more than welcome to continue to use it right so you can uh you know I can create you know I can create a new notebook uh I can make a copy of this modify this and run it and all of that right so all of this is uh and that's why I mean sometimes people say hey that's on the pricer side because I mean uh you know uh we are uh we pay a lot of attention to detail and there's uh of course I mean uh there is there's a lot of overhead on our side that actually um that makes sure that your uh your experience at the boot camp is very very smooth we have we have uh thought through pretty much every single aspect of what makes a a very smooth learning experience so Barry you have a question after the April 22nd boot camp what is the next inperson date in Seattle usually our plan is to uh continue every every two months we have not announced we are still debating you know between you know uh DC Austin and New York and Seattle uh when where should the next one happen so uh please say stay connected and we should be announcing it shortly okay are there any other questions here yes there are uh how long do you have do we have access to the space after the course ends um um legally we tell people uh because we are supposed to you know th