intermediate⏱60 minLesson 6 of 19

Domain 1.6 — AI/ML and Application Development Features

Explore Snowflake's AI and application development ecosystem: Snowpark, Snowflake Cortex (LLM functions, Cortex Search, Cortex Analyst), Snowflake ML, Streamlit, and Notebooks.

Domain 1.6 — AI/ML and Application Development Features

Exam Weight

Domain 1.0 accounts for ~31% of the exam. AI/ML features are a growing focus in the COF-C03 exam reflecting Snowflake's "AI Data Cloud" positioning.

ℹ️Note

This lesson maps to Exam Objective 1.6: Explain AI/ML and application development features, including Snowflake Notebooks, Streamlit, Snowpark, Snowflake Cortex, and Snowflake ML.

💡 Tip: Use Ctrl + Mouse Wheel to zoom, or click the fullscreen icon

Snowpark

Snowpark is Snowflake's developer framework that allows you to write data pipelines and transformations in Python, Java, or Scala using a DataFrame API — without moving data out of Snowflake.

Key Snowpark Concepts

Code is written in Python/Java/Scala using the Snowpark library
Execution happens inside Snowflake — data never leaves
Uses lazy evaluation — operations are built as a query plan and executed when an action is triggered
Supports User-Defined Functions (UDFs), User-Defined Table Functions (UDTFs), and Stored Procedures

python

# Python Snowpark example
from snowflake.snowpark import Session
from snowflake.snowpark.functions import col, sum as snow_sum

# Create a session
session = Session.builder.configs({
    "account": "myaccount",
    "user": "myuser",
    "password": "mypassword",
    "warehouse": "WH_DS",
    "database": "ANALYTICS",
    "schema": "PUBLIC"
}).create()

# Build a DataFrame (no data moved yet — lazy evaluation)
df = session.table("orders")

# Transform
result = (df
    .filter(col("status") == "COMPLETED")
    .group_by("region")
    .agg(snow_sum("amount").alias("total_revenue"))
    .sort("total_revenue", ascending=False)
)

# Execute and show results (triggers computation)
result.show()

# Write results back to Snowflake
result.write.mode("overwrite").save_as_table("revenue_by_region")

Snowpark UDFs and UDTFs

python

# Register a Python UDF in Snowflake
from snowflake.snowpark.functions import udf
from snowflake.snowpark.types import StringType, FloatType

@udf(return_type=FloatType(), input_types=[StringType()])
def sentiment_score(text: str) -> float:
    # This runs inside Snowflake's Python sandbox
    from textblob import TextBlob
    return TextBlob(text).sentiment.polarity

# Use the UDF in a query
df.select(sentiment_score(col("review_text")).alias("sentiment")).show()

Snowpark for Machine Learning

python

from snowflake.ml.modeling.linear_model import LinearRegression
from snowflake.ml.modeling.preprocessing import StandardScaler

# Train a model using Snowflake ML
scaler = StandardScaler(input_cols=["age", "income"], output_cols=["age_scaled", "income_scaled"])
df_scaled = scaler.fit(df).transform(df)

model = LinearRegression(input_cols=["age_scaled", "income_scaled"], label_cols=["churn"])
model.fit(df_scaled)

# Deploy model as UDF inside Snowflake

Snowflake Cortex — AI SQL Functions

Snowflake Cortex provides LLM-powered SQL functions that run directly on Snowflake data — no external API calls required from the user's perspective.

Cortex AI SQL Functions

Function	Description	Example Use Case
`COMPLETE()`	Generate text with an LLM	Summarize, classify, answer questions
`EMBED_TEXT_768()` / `EMBED_TEXT_1024()`	Generate embeddings	Semantic search, similarity
`CLASSIFY_TEXT()`	Classify text into categories	Sentiment, topic classification
`EXTRACT_ANSWER()`	Extract an answer from context text	Q&A from documents
`SENTIMENT()`	Return sentiment score (-1 to 1)	Product review analysis
`SUMMARIZE()`	Summarize long text	News articles, documents
`TRANSLATE()`	Translate text between languages	Multilingual data
`PARSE_DOCUMENT()`	Extract text from PDFs/images	Document processing

sql

-- Summarize customer reviews
SELECT
    review_id,
    SNOWFLAKE.CORTEX.SUMMARIZE(review_text) AS summary
FROM customer_reviews;

-- Sentiment analysis
SELECT
    review_id,
    SNOWFLAKE.CORTEX.SENTIMENT(review_text) AS sentiment_score
FROM customer_reviews;

-- Classify support tickets
SELECT
    ticket_id,
    SNOWFLAKE.CORTEX.CLASSIFY_TEXT(
        ticket_body,
        ['billing', 'technical', 'account', 'general']
    ):label::STRING AS category
FROM support_tickets;

-- Generate completions with LLM
SELECT
    SNOWFLAKE.CORTEX.COMPLETE(
        'mistral-7b',   -- or 'llama3-70b', 'mixtral-8x7b', 'snowflake-arctic'
        'Summarize the following in one sentence: ' || description
    ) AS ai_summary
FROM products;

Cortex Search

Cortex Search provides semantic (vector) search capabilities over Snowflake data without managing embeddings manually:

sql

-- Create a Cortex Search service
CREATE CORTEX SEARCH SERVICE product_search
    ON COLUMN product_description
    WAREHOUSE = WH_SEARCH
    TARGET_LAG = '1 hour'
AS (
    SELECT product_id, product_name, product_description
    FROM products
    WHERE is_active = TRUE
);

-- Query the search service (via API or SQL function)
SELECT SNOWFLAKE.CORTEX.SEARCH_PREVIEW(
    'product_search',
    '{"query": "wireless noise-cancelling headphones", "limit": 5}'
);

Cortex Analyst

Cortex Analyst enables natural language to SQL generation — business users ask questions in plain English and Cortex Analyst returns SQL queries and results:

Built on LLMs fine-tuned for SQL generation
Understands your Snowflake schema and semantic model
Powers conversational BI interfaces
Accessed via REST API or Streamlit integration

Snowflake ML

Snowflake ML is a suite of ML capabilities built into Snowflake:

Feature Store

Store, manage, and share ML features as Snowflake objects:

python

from snowflake.ml.feature_store import FeatureStore, Entity, FeatureView

fs = FeatureStore(session=session, database="ML_DB", name="MY_FEATURE_STORE", ...)

# Define an entity
customer_entity = Entity(name="customer", join_keys=["customer_id"])

# Create a feature view
fv = FeatureView(
    name="customer_features",
    entities=[customer_entity],
    feature_df=df_features,
    refresh_freq="1 day"
)
fs.register_feature_view(fv, version="v1")

Model Registry

Store, version, and deploy ML models inside Snowflake:

python

from snowflake.ml.registry import Registry

reg = Registry(session=session, database_name="ML_DB", schema_name="PUBLIC")

# Log a model
mv = reg.log_model(
    model=trained_sklearn_model,
    model_name="churn_predictor",
    version_name="v1",
    sample_input_data=df_sample
)

# Run inference
predictions = mv.run(df_new_customers, function_name="predict")

AutoML with Snowflake ML

python

from snowflake.ml.modeling.linear_model import LogisticRegression
from snowflake.ml.modeling.model_selection import GridSearchCV

# Cross-validate hyperparameters inside Snowflake
param_grid = {"C": [0.1, 1, 10], "max_iter": [100, 200]}
cv = GridSearchCV(estimator=LogisticRegression(), param_grid=param_grid, cv=5)
cv.fit(df_train)

Snowflake Notebooks

Snowflake Notebooks are Jupyter-style interactive notebooks embedded directly in Snowsight:

Supports SQL, Python (Snowpark), and Markdown cells in the same notebook
Python cells run inside Snowflake — data never leaves
Access Snowflake data directly without connection setup
Version-controlled via Git integration
Can visualize results with popular Python libraries (matplotlib, plotly, altair)

python

# In a Snowflake Notebook Python cell
import streamlit as st
import matplotlib.pyplot as plt

# Load data using Snowpark (already connected)
df = session.table("orders").to_pandas()

# Visualize
fig, ax = plt.subplots()
df.groupby("region")["amount"].sum().plot(kind="bar", ax=ax)
st.pyplot(fig)

sql

-- In a SQL cell within the same notebook
SELECT region, count(*) as order_count
FROM orders
WHERE order_date >= DATEADD('month', -3, CURRENT_DATE)
GROUP BY 1
ORDER BY 2 DESC;

Streamlit in Snowflake

Streamlit in Snowflake allows you to deploy Streamlit Python applications directly inside Snowflake — no external hosting required:

Python data apps run natively in Snowflake
Access Snowflake data securely without API keys
Share with users inside your Snowflake account
Use Snowflake RBAC to control who can access the app
Build dashboards, data explorers, AI-powered apps

python

# streamlit_app.py — runs inside Snowflake
import streamlit as st
from snowflake.snowpark.context import get_active_session

# Get the active Snowflake session (pre-authenticated)
session = get_active_session()

st.title("Sales Dashboard")

# Query Snowflake data
df = session.sql("SELECT region, sum(amount) as revenue FROM orders GROUP BY 1").to_pandas()

st.bar_chart(df.set_index("REGION")["REVENUE"])

# Use Cortex for AI features
query = st.text_input("Ask a question about your data:")
if query:
    response = session.sql(f"""
        SELECT SNOWFLAKE.CORTEX.COMPLETE('mistral-7b', '{query}')
    """).collect()[0][0]
    st.write(response)

sql

-- Deploy a Streamlit app
CREATE STREAMLIT my_dashboard
    ROOT_LOCATION = '@my_stage/streamlit_app'
    MAIN_FILE = 'streamlit_app.py'
    QUERY_WAREHOUSE = WH_BI;

AI/ML Feature Summary Table

Feature	What It Does	Where It Runs
Snowpark	DataFrame API for Python/Java/Scala	Inside Snowflake (compute layer)
Cortex COMPLETE()	LLM text generation	Snowflake-managed LLM endpoints
Cortex SENTIMENT()	Sentiment analysis	Snowflake-managed LLM endpoints
Cortex SUMMARIZE()	Text summarization	Snowflake-managed LLM endpoints
Cortex Search	Semantic search service	Snowflake-managed vector index
Cortex Analyst	NL → SQL conversion	Snowflake-managed LLM
Snowflake ML	Sklearn-compatible ML in Snowflake	Snowflake compute
Feature Store	ML feature management	Snowflake storage + compute
Model Registry	ML model versioning & deployment	Snowflake storage + compute
Notebooks	Interactive SQL + Python IDE	Snowflake compute
Streamlit in Snowflake	Python web app hosting	Snowflake compute

Practice Questions

Q1. A data scientist wants to train a machine learning model using Python without moving data out of Snowflake. Which feature enables this?

A) External functions
B) Snowpark ✅
C) Snowsight Dashboards
D) COPY INTO

Q2. Which Snowflake Cortex function would you use to return a sentiment score between -1 and 1 for customer reviews?

A) COMPLETE()
B) SUMMARIZE()
C) SENTIMENT() ✅
D) EXTRACT_ANSWER()

Q3. A business analyst wants to ask "What were the top 5 products by revenue last quarter?" in plain English and get a SQL query back. Which Cortex feature supports this?

A) Cortex Search
B) Cortex Analyst ✅
C) Cortex COMPLETE()
D) Snowflake ML Registry

Q4. Streamlit in Snowflake applications are secured using which Snowflake mechanism?

A) API keys in the app code
B) Snowflake Role-Based Access Control (RBAC) ✅
C) Third-party OAuth only
D) Public internet access without authentication

Q5. Snowpark uses lazy evaluation. What triggers actual computation?

A) Creating the DataFrame object
B) Calling .filter() or .group_by()
C) Calling an action like .show(), .collect(), or .write ✅
D) Connecting to the session

✅Success

Key Takeaways for Exam Day:

Snowpark = DataFrame API for Python/Java/Scala running INSIDE Snowflake
Cortex SQL functions = LLM capabilities as SQL (SENTIMENT, SUMMARIZE, COMPLETE, etc.)
Cortex Analyst = natural language → SQL for business users
Cortex Search = managed semantic/vector search service
Streamlit in Snowflake = Python web apps hosted and secured within Snowflake
Snowflake Notebooks = Jupyter-style notebooks with SQL + Python cells in Snowsight

Progress32%

Domain 1.5 — Snowflake Storage Concepts

Domain 2.1 — Snowflake Security Model and Principles

Course Content

1. Domain 1.1 — Snowflake Architecture 2. Domain 1.2 — Snowflake Interfaces and Tools 3. Domain 1.3 — Snowflake Object Hierarchy and Types 4. Domain 1.4 — Configuring Virtual Warehouses 5. Domain 1.5 — Snowflake Storage Concepts 6. Domain 1.6 — AI/ML and Application Development Features 7. Domain 2.1 — Snowflake Security Model and Principles 8. Domain 2.2 — Data Governance Features 9. Domain 2.3 — Monitoring and Cost Management 10. Domain 3.1 — Data Loading and Unloading 11. Domain 3.2 — Automated Data Ingestion 12. Domain 3.3 — Snowflake Connectors and Integrations 13. Domain 4.1 — Evaluating Query Performance 14. Domain 4.2 — Optimizing Query Performance 15. Domain 4.3 — Snowflake Caching 16. Domain 4.4 — Data Transformation in Snowflake 17. Domain 5.1 — Data Collaboration, Replication, and Business Continuity 18. Domain 5.2 — Secure Data Sharing and Data Clean Rooms 19. Domain 5.3 — Snowflake Marketplace and Native Apps