I'm building a RAG pipeline and want the simplest way to generate and query embeddings

The simplest practical setup is:

1. Chunk your documents
2. Generate embeddings
3. Store vectors in a vector database
4. Embed the user query
5. Do nearest-neighbor search
6. Pass top matches to the LLM

Minimal stack options

Easiest local prototype

• Embedding model: sentence-transformers
• Vector store: FAISS
• RAG framework: optional LangChain or LlamaIndex

Easiest hosted setup

• Embedding model/API: OpenAI embeddings
• Vector store: Pinecone / Weaviate / Supabase / pgvector
• Or use a managed DB if you already have Postgres

---

Simplest Python example with SentenceTransformers + FAISS

Install

pip install sentence-transformers faiss-cpu numpy

Build index

from sentence_transformers import SentenceTransformer
import faiss
import numpy as np

docs = [
    "RAG stands for retrieval augmented generation.",
    "Embeddings convert text into vectors.",
    "FAISS is a vector search library."
]

model = SentenceTransformer("all-MiniLM-L6-v2")

doc_embeddings = model.encode(docs, convert_to_numpy=True)
doc_embeddings = doc_embeddings.astype("float32")

index = faiss.IndexFlatL2(doc_embeddings.shape[1])
index.add(doc_embeddings)

Query

query = "What is RAG?"
query_embedding = model.encode([query], convert_to_numpy=True).astype("float32")

k = 2
distances, indices = index.search(query_embedding, k)

results = [docs[i] for i in indices[0]]
print(results)

---

If you want the absolute simplest hosted API approach

Use:

• OpenAI embeddings
• pgvector in Postgres

Flow:

• embed text with API
• store vector in pgvector
• query by embedding similarity

This is usually the cleanest “production-lite” option.

---

Recommendation

If you want the fastest path to working RAG:

• Local dev: sentence-transformers + FAISS
• Production: OpenAI embeddings + pgvector or Pinecone

If you want, I can give you:

1. a full minimal RAG script,
2. a LangChain version, or
3. a FastAPI service for embedding + retrieval.