Build and Deploy an AI Chatbot Using LLMs, Python, RunPod, Hugging Face, and React Native
Вставка
- Опубліковано 9 лют 2025
- RunPod: rebrand.ly/Run...
🚀 Introduction:
================================
In this tutorial, we’ll build and deploy a complete coffee shop customer service AI chatbot that takes orders, provides menu info, blocks irrelevant conversations, and even recommends products based on Market Basket Analysis! We'll cover cutting-edge topics like Prompt Engineering, Retrieval-Augmented Generation (RAG), and the modular power of Agent-Based Systems. You'll also learn how to deploy Large Language Models (LLMs) and custom APIs using RunPod, and build a full React Native app that connects to Firebase and the RunPod endpoints. By the end, you’ll have a fully functional chatbot app and level up your AI, development, and deployment skills!
🔄 Update:
Runpod seems to have changed the place to access the openAI URL
here it is: api.runpod.ai/v2/{RUNPOD_ENDPOINT_ID}/openai/v1
and the RunPod endpoint ID is the string right under the endpoint name.
here is a link for more information: docs.runpod.io...
💡 What You’ll Learn:
================================
1. 🧠 Prompt Engineering: Guide your chatbot with precise instructions.
2. 🔍 Retrieval-Augmented Generation (RAG): Enhance chatbot answers using personalized data.
3. 🛠️ Agent-Based Systems: Create specialized components for efficient and accurate chatbot responses.
4. 📊 Market Basket Analysis Recommendation Engine: Build a recommendation engine from scratch.
5. 🖥️ RunPod Deployment: Deploy LLMs, embedding models, and custom APIs effortlessly.
6. 📱 React Native App: Build an end-to-end mobile app connected to Firebase and RunPod.
🔗 Links:
================================
RunPod: rebrand.ly/Run...
Github Repo: github.com/abd...
Coffee Shop Transactions Kaggle Dataset Link: www.kaggle.com...
🎁 Free Credit Codes:
================================
Grab one of the 20 credit codes in the description to get free credits on RunPod!
How to Redeem: Go to the left panel, click on Billing, scroll down to Credit Codes, and paste one of the codes below.
1602zubevdxd7xbzm4ap
mpbictqmksolp73td4mq
opruk1yoqatfc1jw2nry
7l6kusubtdy3cb95906t
7rhjrmch9ilvnwd3dt0r
oiykzqwrk2vhqgkvyh8c
4s5vjcl2irojl1bnkh89
vn7wpd7jkpdnamq3q516
6st9nt72etun8xcvlb6j
svsg0g0fjiuozkaam82t
8kjapravfr1se22126it
6itba529k8083pm15dtj
oy9k1wombmml0pyoo1ba
vyoryb2v9q4tr58etfjh
v6smwvna8c10racrv5si
🔑 TIMESTAMPS
================================
0:00 - Introduction
5:27 - Deploy Llama LLM with RunPod
30:15 - Prompt Engineering Tutorial
52:00 - RAG Introduction
1:15:35 - Recommendation engine Development
2:17:20 - Firebase DB setup
2:47:10 - Pinecone Vector DB setup
3:13:13 - Agent Based System
5:49:20 - Deploy chatbot API with RunPod
6:15:31 - React Native application Front End
11:14:30 - ChatBot React Native Page
🔄 Update:
Runpod seems to have changed the place to access the openAI URL
here it is: api.runpod.ai/v2/{RUNPOD_ENDPOINT_ID}/openai/v1
and the RunPod endpoint ID is the string right under the endpoint name.
here is a link for more information: docs.runpod.io/serverless/workers/vllm/openai-compatibility
If you face any trouble please reach out via email, or Linkedin.
Redeem code is not working none of them
To use firebase storage is not free now. But when you use it at the video it is free.
Broo, 12hours!!! It gonna take me a month. Great contect. Thank you!!!
Love you jiffy just had an order from a client for exactly this product you're helping me so much
did you all the steps in the video ? does it work ?
a lot of thanks. This video will be g.o.a.t. of ai videos. Now, You are my best youtubers.
Glad you enjoyed it! I'll try to keep them coming 😁
I've been looking for something like this. Thank you man. I wanted to know how really to combine those AI systems in apps . Thanks once again
Love it that you use react native for this
You do an amazing job explaining concepts and showcasing great projects. Keep up the fantastic work!
Man I love you! I just love you! Just know, there are people who loves. Thanks for the content!
Bro I have questions.
Does it work?
How much has he spend ?(api and stuff)
Can we make this on our own by doing what he is doing ?
😢
ur a savior sirr......ur the best keep coming with this best contents
This is awesome! Thanks for the tutorial
Ok, bro. You won! I subrscribed already.
thx so much for this tutorial.
This is amazing!!!
Thanks for explain the basic knowledge of concept first , this is mean for me than every tutorial out there because this will help me to deep understand how it really work and apply to others ❤ . BIG RESPECT for u sir
but how much we have to pay in runpod? and also can you give me your github please
@@tarifahmed4956 you can find the pricing in their offical website , run a ai model is exprensive , i choose google colab for learning purpose
Thanks a lot for those kind words 😃
I think 3$ is enough, and the Github is in the description bellow the video.
@@codeinajiffy hope you can make another tutorial like to build a personal langague learning , i am currently work on it without dont know much about llm world , really appricate if u do this thanks
Amazing 😍👏👏👏👏👏👏
Great effort in creating this application! However, I recommend learning the Rasa framework-it would be a better choice for building chatbots with more control
I love you, too! Thank you for the content!
Thank you !!
yo legit tut. amazing
Thank you for building this tutorial. What about in the RAG preprocessing stage to clean data, redact PII information before sending to LLM for chunk enrichment to improve accuracy
Never built a system close to the capabilities of this, but I have an idea for an app, so it's getting done.
Please turn on subtitles for this video. Thank you !!
Hey bro amazing video, your numerator and denominator in the lift formula looks switched as if Support (AUB) is in the denominator then the lift measures how much less likely X and Y are to occur together compared to their independent probabilities.
I will make this project for sure, I promise, 😇😇
Amazzinng... this is so cool
Great work
Hello , I only know python should I got for this project??
Amazing 😍
Great !🔥
OutStanding😍😍😍😍😍😍, Please Please Make Unique LLM's Project , Frontend could be any Language but FastAPI for Model Integration and AWS for Deployment. Waiting for Response and Project Aggressively
Is this ai chatbot project not unique? Please reply. I'm a beginner
@@cutieeeeeeeee-d5j it's Unique but The Role of React Native is too much, But Still It's Very GuDd Project
@@sheikhfurqan3543 Could you explain by what you mean "too much?" Is it like bad or good or difficult? I want to do this as final year project...is this better than the "Football analysis system" this channel owner made? Please help!
@@cutieeeeeeeee-d5j It's a good Project for FYP , because This Full Stack Project has Everything Covered in Single VeDiO
@@sheikhfurqan3543 Is it better than the football analysis one? Can you rate this full stack on a scale of 1-10 in terms of complexity?
Thanks btw! My uni wants our projects to be unique so I'm confused if this can be considered as a "unique" FYP :(
Please make more videos on chatbots using llm.. till deployment
Great project....🎉can you make the next tutorial chatbot that take natural question and respond with sql code where we can query the database with it by using RAG and LLMs it will be a greate content broooo❤
Multi vendor e-commerce app with ai features
sir do projects using cloud platforms like AWS end to end
Hey , but what's the point when we have to go in cart and press on order
amazing can you do in react.js for browser along with one to one chat support and the chat bot option as well
Nice❤
Can you make an AI Course generator for your next project? Please? adding gamified features and other unique things would be so cool! pleasee
Failed to pull image "no matching manifest for linux/amd64 in the manifest list entries"
Getting the above error in runpod logs when docker container in runpod.
Hey I am a student looking to build this for my resume, can someone tell me if later in this video there's a point where I have to make payments somewhere
Do you have this project's code uploaded on git or somewhere else where we can access it
yes, you'll find the link the description bellow the video.
What are the PREREQUISITES I need to have to go through this course ? I have a basic understanding of Supervised ML, Python, React, SQL etc. But no knowledge of Gen AI, Llama, RAG etc. (other than occasionally using ChatGPT).
Nice, you have most of the prerequisites. You need to Basic understanding of programming, python, react and LLMs.
For LLMs, I would say using chat GPT is good step forward. But let's also understand some basics of LLMs quickly. I have a video called Build a text classifier with transformers in 5 minutes. It goes through basic knowledge of LLMs.
how did you record that you could give a break and continue from there since is 12 hours , can you kindly say what tool did you use for recording your lecture ?
I used a free windows software called ShareX. It just records my screen. I can stop the recording and create a new recording and continue from where I left. All manual work.
Sir i really like this project. I do not know how to use React Native. I will deal with it later.
@@muhammadadilnaeem I would say just search how to setup a react project and set it up. And then follow along with the tutorial. Even if you don't understand everything, it will help you to understand what it takes to build an app, and break fears that some people have about not having what it takes to build an app.
Then you can go back watch a react tutorial and you'll find yourself Linking stuff that you coded with knowledge and soon be able to do your own project without any guidance.
But throwing yourself out there and then understanding while you go along is the best way for me to learn anyways.
@@codeinajiffy That's a good Approach Sir. I will give it a try.
pip install-r requirtments.txt didn't work for me did you create a virtual environment prior downloading the required tools ?
Yes I do have a virtual environment but it's not required. Try installing one library at a time. pip install pandas for example and go through all the libraries see if that works
Not able to find Open AI base url on runpod. Anyone knows how to find it?
😢
If you find it please tell how you find it
same here!
Yes Runpod seems to have changed that.
api.runpod.ai/v2/{RUNPOD_ENDPOINT_ID}/openai/v1
and the RunPod endpoint is the string right under the endpoint name.
here is a link for more information: docs.runpod.io/serverless/workers/vllm/openai-compatibility
If you are having more troubles please feel free to send me on the email or Linkedin and having an initial message.
hey was following the tutorial but was disheartened to hear that firebase storage is no longer accessible in the Spark Plan (free) . i wish i could follow the tutorial for free without having to pay for firebase too .
not finding openai base url
Yes Runpod seems to have changed that.
api.runpod.ai/v2/{RUNPOD_ENDPOINT_ID}/openai/v1
and the RunPod endpoint is the string right under the endpoint name.
here is a link for more information: docs.runpod.io/serverless/workers/vllm/openai-compatibility
If you are having more troubles please feel free to send me on the email or Linkedin and having an initial message.
@@codeinajiffy Love Abdullah your work. I cannot support to buy the runpod paid version. Can you please send me any credit access code if you have any remaining. 2ndly any guidelines to use free Groq api key for llama and deploying the chatbot on aws or locally.
Hi ,Can you please make step by step tutorials for fine tuning llms ,please
The video before that, covers that. It fine tunes an LLM for text classification and another for QA
TV seies analysis system video?
@@dozeoffix2255 yes that's the one
what is the alternate of run pod for this app
how did you create your virtual environment can you make a video about that ?
python -m venv my_venv
I have a normal python venv virtual environment but it shouldn't be required
Hai i want to know like our company data is in postgesql and i want to embed that data. How can we achive this? Like through sql connector or something? So if new data gets added to the database how to automate embeding for new data?
A simple solution will be just a scheduled job that runs every hour or so that does embedding on any new data that you have in your database.
how much language supported in this is model?. Can you advise me latest smarter best multilanguage ai model?
Please turn on subtitles. Thanks
Hey is learning and builing apps with Ai and Ml are profitable in this era . Leading to more chances of getting Jobs. Or can we do fullstack without Ai to get jobs.??
Can you make an AI Course generator for your next project? Please?
Is this project open source ?
if no where do I have to spend my money ?
Jiffy, when are you dropping the next project and if possible, can you please give the hint by telling tech stack of the next project.
Hello , I only know python should I got for this project??
@OMINITRIX-x9j Never brother
@@OMINITRIX-x9j but you can try because he has explained the concepts too, the thing is it will be very difficult for you to debug something or tackle the error
@@Ritz7sh what should I learn before coding this
@@Ritz7sh or what u have learn , I want to data science too but I have started mern stack , I just want to build a project for my semester, can you recommend one.?
Hey! Really love the video. Small issue however, none of those credit codes are working for runpod! Please do fix it :D
Hey there, sorry about that people got to them already. I'll see if I can get more codes for you guys
@@codeinajiffy Thank you so much! Keep Growing!
hey bhai how much we have to pay in runpod? credit codes are notr working
Hello, what do i do if i don't have the credits? i mean, runpod is to just deploy? i am making an college project and i wanted to do this but i dont have the credits
bro have you found any solution?
@@tarifahmed4956 no bro sorry :( i had to do another project for my homework
@@MrNebul-ru5fb May I know what project you worked on if there is any link or github repo can you share?
Hello , I only know python should I got for this project??
How do you speed up the chatbot responses?
In the video many chatbot responses are sped up with video edit to not make the viewer wait a long time. But you can speed up responses by either getting a more powerful GPU or using a smaller model like Llama 3.2 1B or 3B parameters. Smaller models should have lesser accuracy and maybe even lesser input context. So it's a trade off between speed, cost and accuracy.
There are other advanced ways to increase model speed like quantization. But that might be for another video.
is this project better than the Football Analysis and Tennis Analysis systems? Please reply. On a scale of 1-10, how unique is this, and will this project land any job?
@@sheikhfurqan3543 nope! from Dubai
@@sheikhfurqan3543 no, dubai
@@sheikhfurqan3543 no
@@sheikhfurqan3543 no. why do you ask?
This project is different from the project you mentioned. This is a natural language processing project while the other two are computer vision projects. This project is totally unique. I will give it a 9. This project will be a powerhouse if you are trying to land a job in jobs that will focus on chatbots as their application.
the runpod codes are not working for me. what to do ?
bro i want to learn these as i am new to this but have full stack developer nextjs and has basic python knowledge background is it possible for me to learn tv series analysis as the first project starting these can that be possible . or i need to start from some other part
There are 2 options:
1- very short way: You can get an API_KEY from OpenAI or Groq kind of sources (Groq gives a limit for free), you can send your prompts and view replies on screen. Model must be Fine-tuned according to your similar aspects. Search it abit. Some guys show it as AI, Deep Learning projects but this is definetely not an AI or Deep Learning. this is just prompt Engineering
2- Long way: You can learn ML, DL then learn finetune a model, then use it. This option is difficult and long, but it gives you unlimited freedom in creating the any content you want.
@@ahmeterdonmez9195 yeah i directly started this till now i am 3 hr in a tv series project and i have been understanding the process. The moment i feel something that i don't get i use chatgpt as simple answer. Hope i could finish it like that if not then i will go long way
But this coffee one i think i would need ml and dl
Hey there, it's awesome that you are understanding this knowledge without any Machine learning background information. What I like to do for fastest knowledge gain is to throw myself into a project and learn along the way just like you. But after I finish I can go and take course on it so everything will stick and the explanation in the courses will be linked with practical knowledge that I already have.
I recommend the NLP course in the deep learning specialization in Coursera.
@@codeinajiffy thanks
Can I use ChatGPT API in this project instead of LLAMA
do you know how much credit we need to add to runpod to full work on this project?
Despite these efforts, the OpenAI Base URL is still not appearing in Runpod.. I got everything you did 😢
same issue did you solved that?
Yes Runpod seems to have changed that.
api.runpod.ai/v2/{RUNPOD_ENDPOINT_ID}/openai/v1
and the RunPod endpoint is the string right under the endpoint name.
here is a link for more information: docs.runpod.io/serverless/workers/vllm/openai-compatibility
If you are having more troubles please feel free to send me on the email or Linkedin and having an initial message.
Yes Runpod seems to have changed that.
api.runpod.ai/v2/{RUNPOD_ENDPOINT_ID}/openai/v1
and the RunPod endpoint is the string right under the endpoint name.
here is a link for more information: docs.runpod.io/serverless/workers/vllm/openai-compatibility
If you are having more troubles please feel free to send me on the email or Linkedin and having an initial message.
Hi Abdul, Nice video.. I am startup founder, I want to talk to you.. what is best way to reach you?
This is not Agentic AI right ?
I love you
Can you build an interior design ai application
❤❤❤❤❤❤❤❤❤
Hi, The codes don't work, how much does it cost to complete this course?
which part of the code that doesn't work ?
What's the issue you are getting? Make sure you have all your environment ready and linked in the .env file.
I think like 3 dollars worth of credit
What to do if we don't have Runpod credits
cant use firebase cause of subscription T_T
Build cab booking app using llms and flutter
I don't know if I am the only one, but the chatbot respond quite slow... So if in real life, I don't think the chatbot with slow response can help people. What do you think? Thank you all.
In the video many chatbot responses are sped up with video edit to not make the viewer wait a long time. But you can speed up responses by either getting a more powerful GPU or using a smaller model like Llama 3.2 1B or 3B parameters. Smaller models should have lesser accuracy and maybe even lesser input context. So it's a trade off between speed, cost and accuracy.
@@codeinajiffy Thanks! Btw, that's a cool project though! ;)
timestamp?
need openai and localdb instead of firebase and supabase instead of pinecone
Timestamps?
If I don't know React Native, can I still watch this? Will it be understandable?
Has anybody tried this project will it work? Please help I am about take it has final year project
Can anyone tell me i am very beginner have no idea on AI tools or applications, can i start this vedio ?
What if we cannot afford RunPod, what else can be an option to atleast learn and practice it for a few weeks?
Your video is GOAT btw! 🥹
Hmm the very first step of my Dockerfile is not working well. Anyone has the same issue? I copy and paste the Dockerfile from Github, and I build the Docker image successfully, and then I run -it --> There is an error at this point saying that: File "/app/agent_controller.py", line 11, in __init__ self.guard_agent = GuardAgent() --> TypeError: __init__() got an unexpected keyword argument 'proxies'
in the worst case, can I just skip that part? Can anyone let me know if I can skip Docker and continue to develop the rest of the application?