Example D: Introduction to RAG with StreamNative Pulsar Functions, OpenAI, and Milvus/Zilliz

In this example we use multiple workflows to create and sync embeddings, complete a similarity search, and summarize the data:

Part 1: Create and Sync Embeddings (OpenAI and Milvus/Zilliz Connector)
Part 2: Similarity Search (OpenAI + Milvus/Zilliz)
Part 3: Summarize Results (AWS Bedrock with Meta’s Llama 3.2)

We will deploy 4 Pulsar Functions and 1 StreamNative Kafka Connector in the following lessons to create the above streaming workflows, testing each functionality as we go. The following setup should be completed before starting.

1. Complete Cluster Setup, Service Account, and pulsarctl.
2. Download code example streamnativerag2.

OpenAI Setup

1. Obtain an API Key for OpenAI.
2. Create a StreamNative secret called myopenaikey with a key called apikey for storing the OpenAI API Key.

Zilliz Setup

1. Create a Zilliz database and collection called mycollection with the following fields.

• primary_key VARCHAR(60)
• vector FLOAT_VECTOR(1536)
• original_text VARCHAR(1000)
• Note that Dynamic Field must be disabled under Advanced Settings for the Milvus Connector to work.

2. Copy the token for your Zilliz cluster and create a StreamNative secret called zillizsecret with a key called mysecret for storing the token. This secret will be used when completing a similarity search on the database. Hang onto a copy of the token. We will need it again when we deploy the Milvus Connector.

AWS Bedrock Setup

1. Enable Meta Llama 3.2 1B Instruct in Amazon Bedrock, model id meta.llama3-2-1b-instruct-v1:0 (note code uses us.meta.llama3-2-1b-instruct-v1:0)
2. Create a user in AWS IAM with the AmazonBedrockFullAccess policy.
3. Create an access key for the IAM user and create a StreamNative secret storing the access key and secret access key.