RAG dengan Cloudflare Vectorize

Cloudflare Vectorize adalah basis data vektor terkelola yang beroperasi di jaringan edge serverless global. Menggabungkan model embedding Neosantara (nusa-embedding-0001) dengan Cloudflare Vectorize memungkinkan eksekusi RAG berlatensi rendah tanpa perlu mengelola server basis data mandiri.

Alur Eksekusi Cepat

import os
import requests
from openai import OpenAI

# Inisialisasi Klien Neosantara
client = OpenAI(
    base_url="https://api.neosantara.xyz/v1",
    api_key=os.environ["NEOSANTARA_API_KEY"]
)

CF_ACCOUNT_ID = os.environ["CLOUDFLARE_ACCOUNT_ID"]
CF_API_TOKEN = os.environ["CLOUDFLARE_API_TOKEN"]
INDEX_NAME = "manual-kendaraan"

cf_headers = {
    "Authorization": f"Bearer {CF_API_TOKEN}",
    "Content-Type": "application/json"
}

# 1. Hasilkan embedding kueri pengguna
query_text = "Bagaimana prosedur pengereman darurat?"
emb_res = client.embeddings.create(
    model="nusa-embedding-0001",
    input=query_text
)
query_vector = emb_res.data[0].embedding

# 2. Query ke Cloudflare Vectorize v2 API
query_url = f"https://api.cloudflare.com/client/v4/accounts/{CF_ACCOUNT_ID}/vectorize/v2/indexes/{INDEX_NAME}/query"
query_payload = {
    "vector": query_vector,
    "topK": 1,
    "returnValues": False,
    "returnMetadata": "all"
}
res = requests.post(query_url, headers=cf_headers, json=query_payload).json()
matched_passage = res["result"]["matches"][0]["metadata"]["text"]

# 3. Grounded generation dengan model chat
completion = client.chat.completions.create(
    model="deepseek-v4.1-flash",
    messages=[
        {"role": "system", "content": f"Jawab berdasarkan data referensi:\n\n{matched_passage}"},
        {"role": "user", "content": query_text}
    ]
)

print(completion.choices[0].message.content)

RAG dengan ChromaDB

Alternatif implementasi RAG lokal dan in-memory menggunakan ChromaDB.

Endpoint Embeddings

Format input dan dimensi vektor model embedding Neosantara.

Langkah Implementasi

Prasyarat Lingkungan

Instal library Python yang dibutuhkan:

pip install -U openai requests

Siapkan variabel lingkungan kredensial:

export NEOSANTARA_API_KEY="nsk_your_api_key"
export CLOUDFLARE_ACCOUNT_ID="your_cloudflare_account_id"
export CLOUDFLARE_API_TOKEN="your_cloudflare_api_token"

Membuat Indeks Vectorize v2

Model nusa-embedding-0001 menghasilkan vektor 768 dimensi. Buat indeks dengan metric cosine:

def create_vectorize_index(account_id, token, index_name, dimension=768):
    url = f"https://api.cloudflare.com/client/v4/accounts/{account_id}/vectorize/v2/indexes"
    headers = {
        "Authorization": f"Bearer {token}",
        "Content-Type": "application/json"
    }
    data = {
        "name": index_name,
        "config": {
            "dimensions": dimension,
            "metric": "cosine"
        }
    }
    r = requests.post(url, headers=headers, json=data)
    if r.status_code in [200, 201]:
        print(f"Indeks '{index_name}' berhasil dibuat.")
    elif r.status_code == 409:
        print(f"Indeks '{index_name}' sudah ada.")
    else:
        r.raise_for_status()

Vektorisasi & Ingest Dokumen (NDJSON)

Cloudflare Vectorize v2 menerima data upload dalam format NDJSON (Newline Delimited JSON). Setiap baris memuat ID, vektor floating-point, dan metadata:

def insert_documents(account_id, token, index_name, documents):
    # 1. Hasilkan embedding via Neosantara
    emb_res = client.embeddings.create(
        model="nusa-embedding-0001",
        input=documents
    )
    
    # 2. Bangun payload NDJSON
    ndjson_lines = []
    for i, doc_text in enumerate(documents):
        record = {
            "id": f"doc_{i}",
            "values": emb_res.data[i].embedding,
            "metadata": {"text": doc_text}
        }
        ndjson_lines.append(json.dumps(record))
    
    ndjson_payload = "\n".join(ndjson_lines)

    # 3. Kirim ke endpoint insert Cloudflare
    url = f"https://api.cloudflare.com/client/v4/accounts/{account_id}/vectorize/v2/indexes/{index_name}/insert"
    headers = {
        "Authorization": f"Bearer {token}",
        "Content-Type": "application/x-ndjson"
    }
    res = requests.post(url, headers=headers, data=ndjson_payload)
    res.raise_for_status()
    print(f"Berhasil mengindeks {len(documents)} dokumen ke Vectorize.")

Pengambilan & Sintesis Respons

Cari kecocokan semantik tertinggi di Vectorize dan gunakan model chat Neosantara untuk menyusun jawaban akhir:

def query_rag(query_text: str):
    # 1. Embed query
    q_emb = client.embeddings.create(
        model="nusa-embedding-0001",
        input=query_text
    ).data[0].embedding

    # 2. Vectorize query
    url = f"https://api.cloudflare.com/client/v4/accounts/{CF_ACCOUNT_ID}/vectorize/v2/indexes/{INDEX_NAME}/query"
    r = requests.post(url, headers=cf_headers, json={
        "vector": q_emb,
        "topK": 2,
        "returnMetadata": "all"
    }).json()

    matches = r.get("result", {}).get("matches", [])
    if not matches:
        return "Informasi tidak ditemukan."

    context = "\n---\n".join([m["metadata"]["text"] for m in matches])

    # 3. Chat completion
    ans = client.chat.completions.create(
        model="deepseek-v4.1-flash",
        messages=[
            {"role": "system", "content": "Jawab pertanyaan secara faktual hanya berdasarkan konteks referensi."},
            {"role": "user", "content": f"Konteks:\n{context}\n\nPertanyaan: {query_text}"}
        ],
        temperature=0.1
    )
    return ans.choices[0].message.content

Contoh Script Produksi Lengkap

rag_vectorize_complete.py

import os
import json
import requests
from openai import OpenAI

NEOSANTARA_API_KEY = os.environ["NEOSANTARA_API_KEY"]
CF_ACCOUNT_ID = os.environ["CLOUDFLARE_ACCOUNT_ID"]
CF_API_TOKEN = os.environ["CLOUDFLARE_API_TOKEN"]
INDEX_NAME = "neosantara-rag-kb"

client = OpenAI(
    base_url="https://api.neosantara.xyz/v1",
    api_key=NEOSANTARA_API_KEY
)

docs = [
    "Ketentuan SLA Layanan: Neosantara menjamin ketersediaan gateway regional Jakarta sebesar 99.95%.",
    "Batas Kuota: Batas rate limit tier Free adalah 10 RPM, 30.000 ITPM, dan 8.000 OTPM.",
    "Pembayaran: Deposit saldo kredit dapat dilakukan melalui QRIS dan Virtual Account Mayar."
]

def main():
    # Buat indeks
    url_create = f"https://api.cloudflare.com/client/v4/accounts/{CF_ACCOUNT_ID}/vectorize/v2/indexes"
    headers = {"Authorization": f"Bearer {CF_API_TOKEN}", "Content-Type": "application/json"}
    requests.post(url_create, headers=headers, json={
        "name": INDEX_NAME,
        "config": {"dimensions": 768, "metric": "cosine"}
    })

    # Ingest data
    emb_res = client.embeddings.create(model="nusa-embedding-0001", input=docs)
    ndjson = "\n".join([
        json.dumps({"id": f"doc_{i}", "values": emb_res.data[i].embedding, "metadata": {"text": t}})
        for i, t in enumerate(docs)
    ])
    url_insert = f"https://api.cloudflare.com/client/v4/accounts/{CF_ACCOUNT_ID}/vectorize/v2/indexes/{INDEX_NAME}/insert"
    requests.post(url_insert, headers={"Authorization": f"Bearer {CF_API_TOKEN}", "Content-Type": "application/x-ndjson"}, data=ndjson)

    # Query
    pertanyaan = "Berapa batas rate limit untuk tier Free di Neosantara?"
    q_vec = client.embeddings.create(model="nusa-embedding-0001", input=pertanyaan).data[0].embedding
    url_query = f"https://api.cloudflare.com/client/v4/accounts/{CF_ACCOUNT_ID}/vectorize/v2/indexes/{INDEX_NAME}/query"
    matches = requests.post(url_query, headers=headers, json={"vector": q_vec, "topK": 1, "returnMetadata": "all"}).json()["result"]["matches"]
    
    passage = matches[0]["metadata"]["text"]
    response = client.chat.completions.create(
        model="deepseek-v4.1-flash",
        messages=[
            {"role": "system", "content": "Jawab pertanyaan secara ringkas berdasarkan referensi."},
            {"role": "user", "content": f"Referensi:\n{passage}\n\nPertanyaan: {pertanyaan}"}
        ]
    )
    print("Jawaban AI:", response.choices[0].message.content)

if __name__ == "__main__":
    main()

Karena Cloudflare Vectorize menyimpan metadata dokumen di dalam setiap vektor record ("metadata": {"text": doc_text}), Anda tidak memerlukan database relasional tambahan untuk memetakan ID vektor ke teks asli dokumen.

Referensi Terkait

Topik terkait

RAG dengan ChromaDB Embeddings Chat Completions Prompt Caching & Optimasi Biaya

⌘I

Panduan

Operasional & Keamanan

Recursive Language Models (RLM)

Teknik AI Lanjutan

RAG dengan Cloudflare Vectorize

Alur Eksekusi Cepat

RAG dengan ChromaDB

Endpoint Embeddings

Langkah Implementasi

Contoh Script Produksi Lengkap

Referensi Terkait

​Alur Eksekusi Cepat

RAG dengan ChromaDB

Endpoint Embeddings

​Langkah Implementasi

​Contoh Script Produksi Lengkap

​Referensi Terkait

Alur Eksekusi Cepat

Langkah Implementasi

Contoh Script Produksi Lengkap

Referensi Terkait