momentoai 0.1.2

Python SDK for Momento AI — image vectorization, face search, and CLIP embeddings with user-supplied Supabase credentials.

🧠 MomentoAI — Python SDK

MomentoAI is a lightweight Python SDK that connects to the hosted MomentoAI FastAPI backend and uses your own Supabase project to store images and embeddings.

Official PyPi published library: https://pypi.org/project/momentoai/

---

📦 Installation

Install MomentoAI via PyPI:

pip install momentoai

---

⚙️ Initialize the Client

from MomentoAI import MomentoAIClient

client = MomentoAIClient(
    api_key="public-access",   # default key for open access
    api_url="https://momento-ai-1-42230574747.asia-south1.run.app",

    # — your own Supabase credentials —
    supabase_url="https://yourproject.supabase.co",
    supabase_service_key="YOUR_SUPABASE_SERVICE_KEY",
    supabase_bucket="YOUR_BUCKET_NAME"
)

---

🩺 Check Backend Status

Use .health() to verify the backend is live:

print(client.health())

Expected output:

{
  "models_loaded": true
}

---

📸 Core SDK Functions

Below are the main operations supported by the MomentoAI SDK.

---

1️⃣ vectorize_image()

Uploads an image → extracts embeddings → stores them in your Supabase project.

response = client.vectorize_image(
    "me.jpg",
    event_id="event1",
    business_id="business1"
)

print(response)

---

2️⃣ find_face()

Finds similar faces in your stored Supabase embeddings.

matches = client.find_face(
    "person.jpg",
    event_id="event1",
    business_id="business1"
)

print(matches)

---

3️⃣ search_images()

Text → Image retrieval using CLIP embeddings.

results = client.search_images(
    "a man smiling outdoors",
    event_id="event1",
    business_id="business1"
)

print(results)

---

4️⃣ list_embeddings()

Retrieve all embeddings for a specific event + business.

records = client.list_embeddings(
    event_id="event1",
    business_id="business1"
)

print(records)

---

5️⃣ delete_embedding()

Delete a specific embedding record from Supabase.

client.delete_embedding("some-embedding-id")

---

🪪 License

MIT License
© 2025 — Manil Modi