Build and Deploy Streamlit App

Streamlit is a Python framework for turning a script into an interactive web app. On dltHub, a Streamlit app is a plain .py file that imports streamlit, and the runtime serves it as an interactive dashboard.

This page walks through building a small dashboard against a loaded dlt dataset and deploying it to app.dlthub.com.

Prerequisites

Add Streamlit to your workspace dependencies:

uv add streamlit

The example below reads from the sample_shop_pipeline that uvx dlthub-start@latest scaffolds. The dashboard needs that data already loaded against the same destination it'll read from:

# Load locally (dev profile, DuckDB) so the dashboard works under `dlthub local serve`
uv run dlthub local run load_sample_shop

# OR, before deploying remotely, load against the prod destination on dltHub
uv run dlthub run load_sample_shop

A dashboard can only display data that's already been loaded. If you skip this step the deployed app boots but every read returns "table not found".

Write the dashboard

Create sample_shop_dashboard.py in the workspace root:

"""Sample shop dashboard."""

import dlt
import streamlit as st


st.set_page_config(page_title="Sample shop", layout="wide")
st.title("Sample shop orders")


@st.cache_data
def load_data():
    dataset = dlt.dataset(destination="warehouse", dataset_name="sample_shop")
    return dataset["orders"].df(), dataset["customers"].df()


orders, customers = load_data()

col1, col2, col3, col4 = st.columns(4)
col1.metric("Orders", f"{len(orders):,}")
col2.metric("Customers", f"{len(customers):,}")
col3.metric("Total revenue", f"${orders['order_total'].sum():,.2f}")
col4.metric("Avg order value", f"${orders['order_total'].mean():.2f}")

st.subheader("Top customers by spend")
by_customer = (
    orders.merge(customers, left_on="customer_id", right_on="id", suffixes=("_o", "_c"))
    .groupby("name", as_index=False)["order_total"].sum()
    .sort_values("order_total", ascending=False)
    .head(10)
)
st.dataframe(
    by_customer.rename(columns={"name": "Customer", "order_total": "Total spend ($)"}),
    width="stretch", hide_index=True,
)

st.subheader("Orders by store")
by_store = orders.groupby("store_id").size().rename("orders").sort_values(ascending=False)
st.bar_chart(by_store)

dlt.dataset(destination, dataset_name) connects directly to a loaded dataset. Wrap the call in @st.cache_data so each widget interaction doesn't requery the destination.

Run it locally

uv run dlthub local serve sample_shop_dashboard.py

This boots the dashboard under the workspace's active local profile (default dev, which reads from .dlt/config.toml) and opens it in your browser.

Configure the access profile

dlthub serve runs interactive jobs under the access profile by default. The minimal scaffold doesn't ship an access profile, so create one before deploying. Add the destination type in .dlt/access.config.toml:

[destination.warehouse]
destination_type = "motherduck"

And the credentials in .dlt/access.secrets.toml. Both database and password need to be in the secrets file for MotherDuck:

[destination.warehouse.credentials]
database = "dlt_test"
password = "<read-only motherduck JWT>"

See the Profiles in dltHub page for the full profile model.

Deploy to dltHub

Add the dashboard to your __deployment__.py manifest so the workspace knows about it:

"""Minimal dltHub workspace."""

from pipeline import load_sample_shop
import sample_shop_dashboard            # module-import → one job

__all__ = ["load_sample_shop", "sample_shop_dashboard"]

Then deploy and serve:

uv run dlthub deploy                                       # publishes manifest + uploads code
uv run dlthub serve sample_shop_dashboard.py               # boots the app remotely, opens URL

dlthub serve runs the app behind the workspace's auth — only your account can open the link. To create a publicly shareable URL:

uv run dlthub job publish sample_shop_dashboard.py         # public URL
uv run dlthub job unpublish sample_shop_dashboard.py       # revoke