Large Language Models Bootcamp- Information Session

Key Takeaways

okay I think we are live um just a minute it shows me that we are live so I think we we probably are live webinars is are now live streaming on custom live streaming service so I will go ahead and get started with sharing my screen here yes we're live now okay thank you SOS um so welcome everyone to the information session for large language models boot camp my name is Raja ibal I am the chief data scientist and one of the lead instructors at data science Dojo um I'm going to talk about some of the large language models boot camp related offerings that we have uh we have different offerings uh um there's an inperson boot camp it's a 5-day 40-hour boot camp that uh we'll look into and then we are also creating uh shorter versions of courses that are instructor-led live uh courses uh interactive live courses but uh in case you cannot make it to the boot camp for you know for various reasons I mean traveling to the location where the boot camp is happening you can still uh attend the boot camp um at within the comfort of your home or your uh office so I'm going to talk about all of those uh and then in general uh answer any questions that all of you may have um we are one of the oldest players in the boot camp uh in data science upscale upscaling space and also one of the most credible players uh in this space right so um very um very focused uh courses we are not purely a training company uh anything that we teach we also do those things right so if we if we are teach teaching large language models we are actually building large language models related products if you talk about fine tuning or rag we are actually doing all of that uh in our day-to-day uh work um 11,000 plus graduates uh more than 3,000 companies globally these are numbers that I mean honestly the numbers have moved on but um Global impact right so we uh have been around for a while and then we are known for the highest uh quality boot camps in Industry um so how did the boot camp actually start right so the boot camp actually started about I would say8 to nine months ago uh when we started uh you know when Chad GPT and Bard and anthropic they started to become mainstream and then companies they started adopting and we started building these solutions for uh companies and um you know we go and use chat GPT uh we ask it a question write a poem for me write a social media post for me uh write uh a letter um to or maybe write an email to a customer who's upset I mean I don't know generate a product plan for me we realized when we started building these uh products for our customers that uh while Chad GPT and Bard and uh other tools out there they make um building these applications uh they make them look quite easy I would say deceptively easy but when you actually um when you actually start building something that has uh you know some Direct business implication and you're uh uh you know you are an Enterprise there are a lot of things that you have to worry about right so well uh Chad GPT may be free but when you deploy an application in your own Enterprise the token token usage and cost they are a consideration then when you talk about um you know uh how big the documents can be I mean does uh or uh do the documents fit in the context window of your models that you're using or not there are regulatory CH challenges I mean easier uh you know it may look uh it may look um yeah I mean what's the big deal but really when you're deploying these uh Tools in uh when there is a business business impact and you are accountable to your share shareholders your customers and the government and the regulatory bodies there are actually some concerns even when you are as an engineer you have to worry about those concerns uh your proprietary data uh how do you guard your data how do you make sure that uh you know your data is not leaked um or your your proprietary um uh data or intellectual property State safe there you have to worry about the inference latencies the data governance and whether you use open source or close Source model um the lack of reproducibility U Can you reproduce the results and when when is it okay to use a model um uh in what scenarios would you use worry about reprodu reproducibility in what scenarios you want I mean hallucination big problem I think even if uh uh you know if you have ever used um one of these tools you already know what Hallucination is uh the knowledge source is changing right it's not uh uh you know you have new customer tickets you have new orders you have new um situations that are arriving how do you actually handle those uh how do you evaluate your models um and how do you design your proms that uh they are not very brittle they don't actually break your application so all of these um these are subtle issues these are uh very interesting fascinating technical issues right so I uh I split my time uh between of course teaching at the boot camp and uh also working um on the other side where we have we have built a product and we have some of the big um you know big companies globally uh that are our customers and they have built uh some llm applications and uh some some tools for these companies they're actively using it in Customer Support uh sales and uh sales marketing education attech uh we are an attech company as well right so we are helping these companies um use these tools that are actually um increasing their productivity by you know 20 30 50 x so uh we have different offerings I will walk you through all of them um very quickly um of course uh the our Flagship product in the llm space is our uh large language models boost boot camp this is an in-person boot camp that happens uh in different cities the next one is coming up starting next Monday on April 22nd uh this is our large language models boot camp this is a 40-hour immersive program so you come in even if you don't have any background in um building laric model applications we are very confident you leave with a working deployed um um you know web application you don't need need to know uh uh any web development you don't have to worry about that so you leave with a working application then we have this uh uh the the idea is that you come in in person uh you start uh within one week There's No context switch right so 8 nine hours a day you spend time you finish the capson project and the next week when you go back to work you're already productive and then then we have this shorter version of courses though for those who cannot make it to the live training the or rather the inperson training then we have these live interactive uh trainings they are shorter duration uh somewhat more affordable and then sometimes say I already know Vector databases what should I do I only am interested in Lang chain what should I do no problem you can always you know pick and choose you can uh boot camp as well the entire boot camp and the live instructor Le courses they are shter courses that you can attend and then we have self-based courses uh uh that I will show you as well um so in the boot camp uh it's uh I have to say we are the only boot camp right so I cannot even say we can we are the best because we are the only boot camp in the world at the moment or we are the first one to actually launch a boot camp uh the first one to Industry and then we have gone to Market um with a uh with actually some very credible Partners right so we have some of the most credible names in industry in the llm uh in generative space they are our partners they are actually involved in teaching this boot camp with us so um and I will show you in a bit about you know the who is involved and what topics they will be involved in uh Hands-On exercises right so this is a very very practical very Hands-On boot camp right so and more Hands-On than many people expect right so even some of the partners they were amazed at when they looked at the curriculum this I mean many of the partners they said I mean yeah I mean we would like to send our own uh Workforce to this because this is comprehensive um and then uh we have a project as well uh these are all of our partners right so if you look at this uh uh actually uh different uh people involved from different companies so we have uh um uh uh in in different areas right so the foundation model the vector databases Etc and then once you uh what we do here is uh when we when you come in um a lot of times U you know there are other hidden costs around when you register for training hey um yes I register for the training but uh I I have to go and sign up for open AI or you know I have to go and sign up for this cloud service what we do here is um you're completely hands off bring in a browser enabled computer computer everything is built in into our sandboxes you have we have browser based coding sandboxes uh you know we will uh will take care of your llm generative AI you know that token consumption cost uh we'll give you a GPU Cloud when we do the fine tuning exercise we have hundreds of maybe more than a thousand code samples uh we do a few dozen during the boot camp and we leave the rest for you so you can be actually working on those uh exercises as you learn um and then also as we improve the curriculum I mean if we have a new talk uh if we have talk or we have a module that we added um this time we started doing that even uh the alumni who have already graduated from our first boot camp we invite them back if we have some interesting speaker so we invite them through a zoom call and they actually can keep up to date so it's it's not that one week and your relationship with us ends we actually keep adding uh adding value to your uh you know your learning and we help you uh with uh you know your growth U even after you have graduated um then we have we have some guest talks uh with real real uh case studies from big companies um as you spend more time in Industry you realize that uh you know technology is only yes technology is important it's a it's a piece in the puzzle but technology should be uh should be uh employed or deployed in in uh in business context right so and we emphasize even our other boot camps other trainings we are known for actually spending time even our deeply technical trainings we actually talk about the business context I mean when to do this and when to do that because at the end of the day anyone can write code anyone can build applications but not everyone can build applications that are scalable that are uh going to cost less are maintainable and all of that so we actually talk about all of those best practices in addition to talking about technology um so discuss various engineering and cultural challenges as I said Hands-On exercise with all topics and then uh at the end of the boot camp building an llm application let me actually walk you through the curriculum here uh let me actually go here on the this course page if you look at this this is our lineup for speakers uh you can see um you know we have um uh we have been adding more people I mean some people may show up this time some people may not right so uh but uh you know the most recent ones that we have added is for instance right so security AI these are our brand new partners and bv8 is also our brand new partner uh we need to update the website here but uh security AI uh security AI is is actually in the business of data in AI governance if you look at this as we start building generative AI applications knowing which dat uh knowing um you know proper data in AI governance is important so I I will explain what that module does so I mean if you look at this this is our uh lineup of speakers and we are constantly adding and uh you know optimizing the list of speakers to bring the best and the brightest from industry and the most experienced people from industry um um so you know it it should not be merely an exercise on running a few python notebooks uh you know there is U there's a knowledge part of things and there's a wisdom part of things and we want to make sure that uh you there are people who are wiser than us we actually hear from them how have they built things that uh have benefited their companies so let me go back and uh talk about this uh and let me go back and talk about this uh uh um the curriculum um 5 day boot camp we start with uh we start with the um the idea of uh the general idea of uh what is uh you know give give you an O overall end to-end overview of what does uh what does the entire ecosystem look like um so two to three hours we kick off the boo camp with the uh giving you the breadth right so before we go dive deep into each of the components we talk about why do you need embeddings and how have embeddings evolved and then we talk about Vector database why do you need a vector database what does you know retrieval mean in a vector dat what does indexing mean in a vector database um what is drag what is fine tuning what are the relative issues what are the risks uh why do you need guard rails um and you know so give you a an end to endend view in two to three hours so we start with that and then we start going and start drilling down deeper into each of the components that you have been exposed to already so you have a you have a bigger understanding of how is an application created then after that when we talk about uh you know we start diving deeper then we start drilling down into embeddings um uh and then we start with the you know the classic way of embedding starting with one hot encoding uh you know talking about tfidf you know count based tfidf uh engrs uh and then word to work and all of that so basically slowly move toward um uh different uh approaches or different uh you know slowly move towards uh the different uh um improvements over the over time in how we used embeddings and then once you have done that you know uh once we have talked about it every theoretical discussion uh is going to be followed by a practical exercise so if you look at this we talked about embeddings we have this practical exercise you know you can see that uh and um and now in this case you can see that this is you can hear you can see all of these um uh all of you anyone who registers they will have a dedicated uh account um in our uh Jupiter notebook and boxes you can modify this lab if you like uh you can make a copy download if you like that's okay with us uh do whatever you need to do and then uh you know run it modify it and play around with this and you have many more exercises you can see that we do some exercises in class and we give some of you some of these exercises for you to follow um uh uh to follow in um on your own when you have time maybe as a homework maybe the next week after the boot camp uh then we talk about attention mechanism and Transformer architecture right so Lis Sano he is the one who teaches it right so he he's you know he goes over then we when we then we do the Hands-On exercise uh then we now that you have the embeddings what do you do with these embeddings how do you store them efficiently then we start with in vv8 is our partner one of the I would say one of the leading Vector databases in the space so so we'll have uh um you know uh one of our uh Partners from VV they will come and talk about you know all the all the challenges or maybe start with how y Vector databas is needed how do you index how do you store and then after that how do you optimize how do you quantize uh for when the when you have too many embeddings how do you um you know how do you optimize your vector database and then same thing right so once you have finished Vector databases it will go and uh and we will give you everything um you know if a vector if it is needed that you have a vector database deployed in uh already that you want to access or if you want to create it this is going to happen you can see that creating a search index we'll go through it loading data into this executing search these these exercises are happening in class by instructors and you're following along right so you can see in the URL uh it is right now you can see there is my name here then you will have your own sandbox and it is going to be you know built in then we actually talk about semantic search I mean basically it's it's the same module uh we talk about prompt engineering uh our prompt engineering exercises uh are there let me show you here uh so you know it's a short module because we don't have a lot of time to spend on prompt engineering we have some deeper uh issues that we need to discuss but uh to give you an idea uh it is there um and we also have some uh really useful self-based learning material uh that you can use for instance we have these prompt engineering sandboxes right so if you would like to practice prompt engineering uh you can see this here I have uh this a separate prompt engineering sandbox I can open it in a new window I'm going to go and copy uh you know um I don't know what this is but you know I just copied and paste it without paying attention but you can see that any prompt engineering um any prompt engineering exercise that you want to do many prompt engineering courses are hey do this do that this is the best practice Chain of Thought you know uh chain of table and all of that um well it's great but over here we are giving you this option for instance if I change the temper what's going to happen so uh and this is included in your uh in your uh in your registration so we actually you you can see that I change the temperature in top p and you can see that it is giving me some gibberish so you get a practical experience it is not some blog or some you know some course that you're reading you are actually you have the ability to go and practice this in the our uh learning environment so that is is there and then we uh get into fine-tuning business uh when we talk about fine-tuning we start with uh you know transfer learning fine-tuning but do you need fine tuning contextualizing it um then we uh go on uh to uh talk about quantization uh low rank adaptation what is uh Laura Q Laura we talk about all of that and then we give you uh everyone gets a dedicated GPU uh uh cluster um thanks to our one of our partners they actually give us credits for that runpod does it and we give you those GPU credits and then as a result of that uh you uh and we also give you um a latu uh model uh that's four bit quantize 7 billion parameter model we give you the data set we walk again instructor guided exercise we walk you through uh this is how we are fetching the model from hugging face then we did this and then we did this and then we did this and then when you are done with it you evaluate the your uh your fine-tune model with a model that was not fine-tune and compare and contrast how the uh responses were um so there is a lot of uh there's a lot of value in it right so attending a course where someone is talking about fine-tuning or actually attending a lecture followed by doing fine tuning yourself I mean definitely I mean reading a book on how to ride a bicycle or how to swim in a swimming pool that's different than actually diving uh taking a plunge and then starting to swim in maybe Open Water right so so I think uh there is no replacement no substitute and we are known for that idea that you know yes I mean Great Courses out there on many of the muks uh but at the end of the day you know things in theory we actually um you know well um you know oh a course on Open Water swimming right so you know you go there you have to worry about this and that right so but actually when you uh you know when you actually experience the experience the frigid water maybe the lake I mean there there going to be boats just there you don't go here don't go there right so there is instructions and there's practical experience the boot camp is about actually tossing you in a lake and let you actually swim and survive so when you go back you're better prepared for real world right so it's it's not a theory course it's a practical course so uh then we have uh domain specific model we talk about uh we talk about um um why do you need a domain specific model so this is uh uh on how do you create conversational model so once again uh we talk about this and we have a Hands-On exercise uh we spent quite a bit of time on Lang chain uh Lang chain um I'm sure if you're here you must have heard of Lang chain already uh Lang chain uh we talk about all the Lang chain components let me actually go and perhaps bring it up here so if you look at this this is our Lang chain module uh we'll bring it out Lang chain for lication development right so so if you look at this all the Lang chain component this is I think this is the longest uh or lengthiest mod module that we have so we start with the you know model Io if you look at this you know prompt templates example selector chat models and the same drill that I showed you you click on it exercise pops up how do you create templates uh you can see that it is asking for an open AI key we give you the open AI Keys We bear the cost of uh uh of that open AI expense you can see that uh uh you know we are using GPT 3.5 turbo plug in the openi key and uh and this also shows you sometimes people sayy I'm not I'm not actively coding as long as you can read code you will be fine you know if you have a use case even if you're a technical product manager if your product manager wants to build products but you have not coded in a while or you're not a coder by training uh you know uh we have a Python tutorial two-hour tutorial that is good enough for you uh to um you know gramp up on python that is needed for this course and after that that's it right so you know um you uh run these uh notebooks run run run and you can see it is not letting me run this notebook because I've not given provided it the uh the API key but once I have provided it the API key it should actually work I mean you can see that it should be actually working if I did that so even if you don't have a coding background but you want to actually learn Li language models this is training the training for you but if you have never coded you don't know what coding means of course I'm not advocating that but if you if you can code in one language maybe you have coded uh you if you can write SQL you generally understand what you're dealing with you will be fine I mean we have had people who did not know coding and they survived uh or shouldn't survive is probably not the best word they actually did quite well in the uh depending upon you know what they expected and if you have any doubts please feel free to set up a time with uh us uh and we are going to happy to help you in um in deciding whether this boot camp is for you or not um fine-tuning uh conversational AI models Langan have talked about it then observability and uh uh and and uh and uh evaluation uh so um in in how do you actually log everything that your model is doing I mean I asked this question I that that question how do you put guard rails around uh around your uh prompts how do you put guard rails around your responses how do you prevent uh your uh your model from giving any information that it should not be giving uh in uh you know you deployed uh you deployed a model let me I think um there's a relevant example Air Canada actually recently deployed this uh model uh that uh uh it was a llm based uh um chat bot customer support bot one of the customers who was visiting a family in a different city maybe from Vancouver to Toronto and uh the chatbot offered bement fairs uh and um well yeah you go ahead said buy a ticket when you come back you will get I don't know 80% or 90% back the customer said very well uh went for attending the funeral came back and um asked call air called Air Canada where is my money and Air Canada says uh what money I mean bement fa we don't offer but some Airlines do Air Canada did not and then whoever developed that application did not understand certain things about how to build the models God rails and you know these kind of uh you know oversight uh on your model uh to prevent it from doing these things that is actually uh that is very relevant right so uh so we actually talk about all of those cases and how do you build applications that are secure how do how do you build applications that actually do not result in uh that do not result in uh you know any adverse business impact for you so the court actually ruled in favor of the customer and then Air Canada had to actually pay and interesting Court ruling um we are we are living in interesting times where new presidents are going to be set up right so if it was your chat bot you are responsible right so you cannot say it was generative Ai and we are not responsible as Air Canada um you owe uh the customer that bement fair fix your chatbot um because your chatbot actually told uh the customer and then pay the customer the that credit um then we have a uh then then we have a a proper module on uh on evaluation we talk about evaluation in detail how do you evaluate um a model um in fraud detection for instance uh you can actually have a model uh if it is a fraud or not a fraud supervised learning yeah I mean evaluation is easy but what if U if I have the response of the chatbot is correct but it is worded differently so there are a lot of uh once again there's a lot of subtleties and lot of uh finer points that we need to be aware of so we talk about evaluation in detail and then once again when I go to evaluation I'm going to look at the Practical exercises there are practical exercises for evaluation so everything I'm not going to go and click on every module but I will go show you this occasionally you can see that we are talking about you know different uh different ways of evaluating and then we are going to actually go and go through this evaluation criteria these notebooks may look intimidating but by the end of you know by by the time we reach there you will already fairly uh understand you you why this looks like this and and so on so I mean we have had great success with people actually ramping up on these uh then uh on uh we also have added um one module that may not be here but it is going to be uh added um what we have built uh we have been building a product that has customers and that product has actually taught a lot and one of the things is uh you know building a basic rag application it is a piece of cake I mean I'm not exaggerating basing building a basic rag application a simple rag application retrieval augmented generation application it's a piece of cake but as you get more complex use cases um you know rag actually becomes very complicated right so you have uh all the way to to you know a simple a single data source a single document to you know a big document repository and then you talk about semantic search and lexical search and how do you the inter interplay between the the two the rank Fusion so we have a a module on Advanced rag application because rag is going to be the um for most companies rag is going to be the uh the Paradigm of choice for building llm applications for most Enterprises so we talk about even though we teach fine tuning but we think that rag is going to be the Paradigm of choice there are issues like Access Control let's say if uh I built uh an application and I uploaded a document and I only I should have access to the document someone else also has the access to the doc uh to the application now should they be able to see the response or not um what about multi-agent type situation and uh we talk about a lot of these subtle issues um that uh that actually can have a a lot of impact in uh in your success uh in this llm adoption Journey um then we also have one of our partners vcta they actually come and talk about um rag in a box they actually in case your Enterprise has certain limitations maybe you are one of those who doesn't uh you know you don't you're not the the technical technology company you just want something that uh you know easy to adopt so they they basically offer a rag uh platform that you can use um so we have a talk uh and a a tutorial and a Hands-On uh exercise from vcta and then we also have security AI we have uh uh security AI they are in data data in AI governance for rag applications so they will be talking about uh that on the last day we will actually learn how to deploy uh operationalize these models or these applications and then in addition to that uh on the last day we are going to actually build um an endtoend application and once again even if you're not a Cod coder we know what we are doing we know exactly um I personally trained SE 8,000 plus people in a face-to-face setting so um so I I can I I consider myself an educator first before anything else so um what we have uh done is we have we have built the entire infrastructure in a way um or built the entire pedagogy or uh the the entire uh the structure of boot camp in a way that even if you're not a coder you should be able to actually uh get by with it so what we do on the last day is we give you uh every attendee gets a VM uh a virtual machine uh and login credentials you log in you already have a web application code uh that is running um uh you have uh you know your um coding ID is already installed um um we walk you through the code and uh you connect it with your GitHub account you connect you create a streamlet account you push your app and you have your first rag application deployed and then we give you exercises depending upon who you are I mean if you are a hardcoree coder um by the end of the training you have tons of ideas you go and modify the function so the basic skeleton of the web application is there streamlet makes it very easy um and then uh we give you exercises can you add this chain to it I mean can you add more documents can you can you create a prompt template uh can you create a Wikipedia or dougd goo or a Google agent right so so we depending upon what level you are at uh in before you came to the boot camp um you know you get different mileage out of it of course but everyone leaves with a working um rag application and um everyone completely understand how it is done and your ability uh to go and build of course it depends upon we are not selling snake oil here right so you know if you are not the best coder we are not teaching coding here but you absolutely understand how application llm applications are built so I'm very confident uh anyone who is in technology can actually finish the course without worrying about the coding skills but whether you will be able to build you know Cutting Edge new applications and new products it will depend upon your prior background right so because we are not teaching you uh the fundamentals of distributed systems here we are not teaching you how to how to create a Lambda or Azure function we are not teaching you the the subtleties of uh you know Cloud infrastructure so but in general if you are here for understanding the llm ecosystem uh as I keep saying right so we are the we the only boot camp so you have to be the best right but you know on that on a serious note we are actually we have done uh given it a lot of thought every boot camp we learn and we have been improving every time um that is it and I have I see a barrage of questions in my chat here let me see uh I will uh I will answer them in the order they arrive and if you are attending a live stream if you're attending a live stream uh please ask the question in whatever Channel you I think we are live streaming this on YouTube and uh LinkedIn or any other place please uh um uh please uh uh post your question there and then it will be routed to me okay um we are are you planning for online boot camp yes we are planning for uh we already have announced some of the boot camps we will not be offering this entire boot camp in a 5day for format online unless uh you know you know I have seen people who can do it but most people they would not want to sit on a zoom call for five days straight okay so but what I can show you here is I can show you what we have where is that let me see think I to open it somewhere here but I can open it again never mind so so if you look at this we have recently announced uh this uh so we are breaking we are breaking this uh this course uh down into uh shorter courses so this is the very introductory course this is uh so this is the entire boot camp the input and boot camp but if you want to finish this uh step by step so this is large language models for everyone what that means is no matter what background you're coming from you can be uh you can be a lawyer you can be you know you can be a technology professional of course don't get me wrong here you can be hardcore technologist you can be someone who has never uh done any technology but you want to know what what is all of this about and we are going to actually teach you you know what uh what is all this Buzz about what what are llms and what are um what are llms and what are what is generative AI what is prompt engineering and how is everything structured right so this is a very short half day uh course uh this is live uh online and you will you know you will have get get access to the resources uh uh to this related this course then mastering Lang chain this is our full day course uh that actually will anyone who wants to anyone anyone who wants to be an uh um LM application developer if you want to build your own rag application Lang chain is a must have skill llama index Lang R Lang chain all of them so uh this is the course this is a full day course uh uh one day uh and the courses are actually very inter interactive you are going to uh you know you can ask questions the same same way the labs that we have uh that I was showing you the same structure it is just the Lang chain course that you have I should have mentioned all of our courses also come with a certificate from uh the University of New Mexico with uh uh credit uh continuing education credit so if your company requires the course to have some some kind of cus for reimbursement process yes we have our certification you can um you can actually request a transcript from the University of New Mexico uh with of course relevant credit so uh you know we have this academic partnership uh with them so this is and we are we will be adding more courses it may look like hey what's a big deal add more courses but I mean each course actually takes an insane amount of uh effort that we are trying to well you know we are trying to add more courses as uh time permits let me me see uh what other are you planning to yeah this is done um then uh laptop screen will not help me hence exploring do will allow me to bring my own ex absolutely Sur bring your own external monitor as long as it can fit in in our classroom uh it should be fine so please uh feel free to right so um I think uh and you don't have to bring your external monitor we may be able to because the the Seattle training is actually happening in our uh in our facility uh and then we have actually we we should be able to actually offer you extra monitor uh if you need to okay um yeah uh I joined late how much is the cost so the cost of the boot camp uh right now it uh it is basically uh based on uh the demand and the Supply right now it is at $ 34.99 uh for the Seattle boot camp so basically depending upon what seats how many seats are remaining and how full the boot camp is um it changes reach out to us and we are happy to actually help you out and guide you with that um to find out uh let me actually go here yeah so um let me go here okay so to where is the other question is there a link for that course yes uh so Redan uh there is a link for the course you can check it out uh you can check it out here it is on our website you can take a look at that uh prerequisites required for this course all you need to do is have some commitment uh some commitment some some drive and maybe a little bit of programming experience we will take care of everything else then uh again the same questione share the prerequisites require skills required can you share info about your python course uh are you referring to the python course that we offer are you talk uh talking about the python prerequisite because the prerequisite python uh we will uh it's part of this and um it is already included and as soon as you sign up we'll send you that link to that tutorial what is the price in schedule for the 4 to8 Hour uh llm short courses I mentioned that um even though the list is not complete uh you can see that we have only one uh two more courses but we have a course coming up with fine-tuning we have a course coming up in front engineering we have a course coming up and observability and uh evaluation and guardrails so we will be uh we will be actually um announcing those as uh we uh proceed uh where is the in-person boot camp located right so we have offered the these boot camps in different locations um but currently uh we are the next boot camp has is happening in Seattle um and we plan to actually offer it more frequently with in Seattle so uh let us know what location you Ruth what location you're considering uh let us know because it takes a uh I I I think I did not mention right so when you are um when we announce a boot camp in a in a different city the biggest challenge is not actually uh the training itself it is the logistics right getting the venue because uh we don't cut Corners in venues I mean you have to have some proper chairs proper tables economic chairs and all of that right so we we need to make sure that I mean it's eight hours a day and we have to make sure that you have hot breakfast Hot Lunch you know coffee and refreshments and all of that during the day so it it takes a bit of uh the logistics tend to be tends to be uh it is a difficult finding a venue for the boot camp we are based in Seattle so it's very easy to offer a boot camp in Seattle for other places we have done uh boot camps uh in 17 different locations globally when it comes to our data science boot camp but uh for llm boot camp we have done it in uh DC and Austin and likely we will go back to DC and Austin and possibly at New York as well but uh not at this point I mean it um there's nothing at this point that we have planned uh but letting you know what we have done in the past and then future will decide uh are there any boot camps in San Francisco Bay Area Ashish if you can convince the state of California to regulate uh their regulatory bodies they are actually too active and we it's very hard to actually deal with their Regulators so we are not going to be doing anything in San Francisco Bay Area unfortunately we know that that is the place to be but the problem is that uh jumping through the regulatory Hoops uh it is not worth the effort so we don't plan to offer anything in California okay uh do you know any good s for AI code assistance for python code when I do not have a computer science background actually as a matter of fact you can actually come and sign up uh this we have some free courses that offer AI code assistance for python if you go to online for online. datascience dojo.com um pick up like uh you know data visualization using python uh click on it uh so there is a course this is a free course uh anyone can sign up histograms and then you can see that uh if I run it is giving me this histogram and now let's say if I let me just add a a index error intentionally here right so if I did that you know it will uh you know it will tell me and then I when I go to the AI tutor see so AI tutor is going to help me and then um you know I can actually uh it is suggesting me some exercises uh it is telling me right a Python program to to create a line uh plot of given array of number so it automatically it is based on what you did just now it is suggesting you can you give me a codee sample for exercise one and uh it is going to go ahead and generate a code sample for the exercise it has created you can see that it is already giving me you know that uh that plot the line plot that it gave me so uh so Kelly um you have it available for free just go ahead and start you all you have to do is just go to online. data science do.com and get started uh uh you don't have to sign up for anything uh to uh to access this will there be office hours for the short course students um um we have uh because they are a single day course we um for our other longer term uh multi multi-week or multi-day courses yes we do have office hours but for these particular courses uh you know it's a 4-Hour course and office hours might be actually too much of a you know operational overhead for us having said that we are one of the most customer Centric companies I mean many people even several years after attending the training if they come in for a a reasonable request we actually accommodate those requests I cannot make a promise but we have we usually go out of the way to help people but sometimes people would say hey I'm doing this work uh some Consulting work and uh you know I have this problem at work can you help me so those kind of things we cannot but anything related to the material we usually are pretty pretty cool with that right so we will be happy to but there are no officially announced office hours for the online the short course students um okay and then I see users on user on YouTube will these videos be available for later usage yes only for the boot camp attendees as a matter of fact uh we actually if there's a new session that as we have added that the previous boot camp attendees did not have we actually make these uh uh the sessions available uh or live sessions available let's say we have a new speaker we think uh that uh uh we we think that uh someone who previously attended the boot camp and now this is a great talk that we have added we will actually go ahead and um uh invite to the you to the live session and then in addition to that you can see that we have uh the recordings from the previous session so uh yes we do have these recordings from previous sessions that are available um and then we have questions in the Q&A tab is there any new module focused on development of multi-agent systems TR yes uh we have it but uh I mean looks like you have been dabbling with this so we do actually talk about multi-agent systems as well but to what degree of detail I want to be honest to you you know because it's it's a 40-hour boot camp you have to cover so many things so we do talk about you know how do you create a multi-agent system in uh in Lang chain but nothing very complex uh you know you you but yes there is a m agent system um when we talk about agents here let me show you where is it l chain agents right so if you look at this so we do talk about it here uh you know you can see that and practical exercise you can see that agents and tools so yes I mean we talk about it and this exercise I know you create a uh you create multiple tools and then it goes and talks to multiple agents and the multi-agent question I asked above is referring during the boot camp absolutely we do it during the boot camp would there be session how to host serve and open llm on a cloud platform like ec2 um CR um no um we are not going to talk about this but the I'm very very confident that after the boot camp um you should be able to go and uh um do it on your own because yes we will talk about all of these but not actually in ec2 because you know there are so many platforms out there unfortunately we cannot cover everything and the way we approach learning is we don't approach learning as or uh our trainings as uh uh you know we we are not here for a checklist we are not there to teach everything our job is to actually inspire you our job is to our job is to really uh enable you and inspire you and give you the the minimum background and once you have that background you should be able to do uh and whatever you need to do and we have a long history of it right so I mean you know I'm sure that many there are many other trainings and people make tall claims go and check out our reviews I mean there's uh I don't think any other company uh in our space has that many actual uh you know verified reviews on the website you know you can go and check out hundreds of people on our website they are leading positions I mean partners and VPS at companies who would uh only say good things about us right so you know we we love what we do so uh but going back to it um you can you do it after the boot camp yes do we teach specifically posting and serving an open llm model on a cloud platform like E2 no we don't specifically teach this but by the end of the boot camp you will be able to do it yourself I I hope I answered the question okay then let me see are there any other question I think there is okay thank you crish um do we have anything left and um I think this is uh maybe at a very high level right so whatever whatever I went through is essentially this right if you look at this I'm sorry I should actually go and talk about it so if you look at this here we we talk about the different components of uh you know we talk about Vector data databases how do you create embeddings how we talk about orchestration or L chain type you know llama index and all all those things we talk about santic caching login deployment uh you know guard rails uh using open source and so uh and closed Source models we use mainly uh we have used U uh we use llama Tu for the fine-tuning exercise and mainly we use GPD 3.5 and 3 uh and gbd4 and then we use streamlet for application so you can see that the entire ecosystem the our this is our reference architecture right so we actually uh we actually would uh teach you the the entire end to endend um you know the bigger picture of this llm ecosystem and not just focus on something very specific right so you will you will leave with a a general solid understanding of the entire uh ecosystem and then there's so many things that of course I cannot cover in a single hour and this is actually the entire curriculum I have already gone through this you can take a look at it uh and uh let me see what else do we have this is our line of instructors um I think Sophie and Hamza are not available this time but rest of them are going to be teaching every so you see this is our line of instructors everyone is going to be there except uh you know two instructors for the Seattle boot camp they are not available this time um past testimonials I mean this these are not some John do and Jane do I mean to go talk to these people these people you know they are uh theyve been involved in the boot camp these are our partners these are our customers and these are actually right now these companies if you look at this these companies have actually attended our training right so we have been trusted by these companies uh you know we have people traveling from all the way from Middle East um um aramco and then all the way from Australia people have traveled to attend this boot camp so you know and I I think that should at least show some credibility there uh that people trust us and then we have uh you know the the next one that is happening in Seattle is two three four days from now it is starting if you're in the Seattle area uh that is going to happen in April and then after that we have another one coming up in June um about two months from now and then uh and then we will be announcing more locations as our schedule permits uh and the online trainings will continuously keep uh uh announcing we have two trainings that are finalized uh on in the online format um and then we will keep adding more of these I'm happy to take any more questions that you may have okay this is wonderful um please reach out uh we have uh shared some of the URLs with you set up a time if you have any questions uh set up a time with us we are happy to guide you help you but uh you I'm looking forward to seeing uh some of you in one of our future boot camps or the online trainings thanks every mon