Building LLM Model BOT with Redpanda

Project Architecture

Inspiration

We are in the era of fast rapid information being shared and transferred in real time. With the help of the Large Language Model human intelligence and performance have improved more than ever before. Ability to solve more complex problems faster and more precisely.

With streaming technology being on the rise we plan to build an LLM chatbot and deploy it to Telegram where users can interact with the solution in real time and get precise responses.

What it does

Step 1: Set up Docker-Compose.yml

This will contain all the different services needed for this project. As we plan to store the data in ElasticSearch and visualize it with Kibana.

Step 2: Create a Cluster and Topic in the Redpanda Site

Create the cluster needed for the project and create a topic to which all the data will be sent.

Step 3: Create both Producer Script and Consumer Script

As this project will be using Python to send data to Redpanda topic consumes the data and sends it to ElasticSearch and Azure Data Lake Gen 2.

Step 4: Data Transformation and Standardization

The stored data in the Azure storage account are then transformed and prepared for training purposes.

How we built it

The technology used is Docker Container, Azure Storage Account, and Python. Docker was used in setting up ElasticSearch and Kibana which serves as the storage for the data consumed from Redpanda Topic. Data consumed from Redpanda is also sent to Data Lake Gen 2 and lastly, we built an LLM model using OpenAI for the data in Azure Data Lake for business users.