20 Snowflake Projects for Beginners and Intermediate Learners

Compare 20 Snowflake Project Ideas

Use this table to choose a project based on your current SQL skills and the output you want to build.

These are learning exercises. Advanced extensions later in this guide explain how to add operational requirements.

Which Snowflake Project Should You Start With?

Snowflake Projects for Beginners

Start with Project 1 if you understand SELECT, WHERE, and GROUP BY but have not built a complete data workflow.

Continue with customer purchases, inventory, invoices, or subscriptions. Focus on checking your results and explaining how missing values, duplicate records, and joins affect the answer.

Snowflake Projects for Intermediate Learners

Choose funnel analysis, customer loading, historical reporting, or batch reconciliation when you are comfortable with joins and window functions.

These projects introduce more demanding decisions about event order, update rules, effective dates, and repeated processing.

Moving Toward Advanced Projects

After validating a foundation project, add file ingestion, scheduling, monitoring, access controls, and recovery.

A working SQL query is the starting point. An operational pipeline also needs evidence that it handles failures and changing input correctly.

Plan Your Project Before You Start

Project duration depends on your experience, account access, dataset, and extensions. Complete one reproducible build before adding more tools.

Required Tools, Account Access, and Datasets

You need access to Snowflake, a SQL worksheet in Snowsight, and permission to create the required practice objects.

Use a dedicated practice schema. Compute and storage may incur charges, so understand your account’s usage arrangements before starting.

Check your worksheet context:

SELECT
    CURRENT_ROLE(),
    CURRENT_WAREHOUSE(),
    CURRENT_DATABASE(),
    CURRENT_SCHEMA();

The first walkthrough uses temporary objects. Run it in the same session and save your scripts before disconnecting.

Python, external cloud storage, orchestration tools, and BI software are optional extensions.

Resources for Snowflake Projects With Datasets

Each project includes SQL examples, synthetic sample inputs, expected results and validation guidance. Complete Project 1 before Projects 2, 5 and 20. Run dependent temporary objects in the same worksheet session. Save each project’s:

• Source records and dataset documentation.
• Setup and loading scripts.
• Transformation and validation queries.
• Reporting queries.
• Actual results and screenshots.
• Assumptions and known limitations.

When using public data, document its source, usage terms, column meanings, and update frequency.

Project 1: A Snowflake End-to-End Project for Sales-Data Quality

Business question: What paid-order value did the business record each day?

Problem: The source export contains duplicate orders, an invalid date, and a negative amount.

Skills: Loading, type conversion, validation, deduplication, aggregation, and reconciliation.

Deliverables: Raw input, rejected records, clean orders, and a daily paid-order report.

Understand the Workflow

This walkthrough covers loading through reporting. External extraction, scheduling, and monitoring can be added later.

Step 1: Create and Load the Dataset

Run the following in your practice schema:

CREATE OR REPLACE TEMPORARY TABLE sp_orders_raw (
    source_row INTEGER,
    order_id INTEGER,
    order_date_raw VARCHAR,
    customer_id VARCHAR,
    amount_raw VARCHAR,
    status VARCHAR
);

INSERT INTO sp_orders_raw VALUES
    (1, 1001, '2026-10-01', 'C01', '1200', 'paid'),
    (2, 1002, '2026-10-01', 'C02', '800', 'paid'),
    (3, 1003, '2026-10-02', 'C01', '650', 'refunded'),
    (4, 1004, '2026-10-02', 'C03', '2000', 'paid'),
    (5, 1004, '2026-10-02', 'C03', '2000', 'paid'),
    (6, 1005, 'not-a-date', 'C04', '900', 'paid'),
    (7, 1006, '2026-10-03', 'C05', '-50', 'paid');

Check the input count:

SELECT COUNT(*) AS raw_rows
FROM sp_orders_raw;

The expected count is seven.

For file-loading practice, save the same records as CSV and load them into a separate practice table. Use either the SQL seed or CSV as the baseline input to avoid loading both.

Step 2: Convert and Validate Values

Require an order ID, customer ID, valid date, nonnegative amount, and supported status.

CREATE OR REPLACE TEMPORARY VIEW sp_orders_typed AS
SELECT
    source_row,
    order_id,
    TRY_TO_DATE(order_date_raw, 'YYYY-MM-DD') AS order_date,
    customer_id,
    TRY_TO_DECIMAL(amount_raw, 12, 2) AS amount,
    LOWER(TRIM(status)) AS status
FROM sp_orders_raw;

CREATE OR REPLACE TEMPORARY VIEW sp_orders_checked AS
SELECT
    *,
    CASE
        WHEN order_id IS NULL THEN 'missing_order_id'
        WHEN NULLIF(TRIM(customer_id), '') IS NULL
            THEN 'missing_customer_id'
        WHEN order_date IS NULL THEN 'invalid_order_date'
        WHEN amount IS NULL THEN 'invalid_amount'
        WHEN amount < 0 THEN 'negative_amount'
        WHEN status IS NULL
            OR status NOT IN ('paid', 'refunded')
            THEN 'unsupported_status'
        ELSE NULL
    END AS rejection_reason
FROM sp_orders_typed;

This rule records the first failure encountered. You can extend it to retain multiple failure reasons.

Step 3: Retain Rejected Records

CREATE OR REPLACE TEMPORARY VIEW sp_orders_rejected AS
SELECT *
FROM sp_orders_checked
WHERE rejection_reason IS NOT NULL;

SELECT
    source_row,
    order_id,
    rejection_reason
FROM sp_orders_rejected
ORDER BY source_row;

Expected output:

Step 4: Remove Duplicate Orders

For this exercise, retain the valid record with the lowest source-row number per order ID.

CREATE OR REPLACE TEMPORARY VIEW sp_orders_clean AS
SELECT
    source_row,
    order_id,
    order_date,
    customer_id,
    amount,
    status
FROM sp_orders_checked
WHERE rejection_reason IS NULL
QUALIFY ROW_NUMBER() OVER (
    PARTITION BY order_id
    ORDER BY source_row
) = 1;

A changing source feed needs an explicit version or event-ordering policy. Document that policy before applying updates.

Step 5: Produce the Report

SELECT
    order_date,
    COUNT(*) AS paid_orders,
    SUM(amount) AS paid_order_value
FROM sp_orders_clean
WHERE status = 'paid'
GROUP BY order_date
ORDER BY order_date;

Expected output:

The refunded order remains in the clean dataset but is excluded from this measure.

Step 6: Reconcile the Results

SELECT
    (SELECT COUNT(*) FROM sp_orders_raw) AS raw_rows,
    (SELECT COUNT(*) FROM sp_orders_rejected) AS rejected_rows,
    (SELECT COUNT(*) FROM sp_orders_clean) AS clean_orders,
    (
        SELECT COUNT(*)
        FROM sp_orders_checked
        WHERE rejection_reason IS NULL
    ) - (
        SELECT COUNT(*)
        FROM sp_orders_clean
    ) AS duplicate_rows_removed;

Expected counts:

Check key uniqueness:

SELECT order_id, COUNT(*) AS records
FROM sp_orders_clean
GROUP BY order_id
HAVING COUNT(*) > 1;

This query should return no rows.

Step 7: Test a Changed Input

Change order 1003 from refunded to paid and rerun the report. Total paid-order value should increase from 4,000 to 4,650.

Record your prediction, actual result, and explanation.

Capture actual outputs from your Snowflake workspace before presenting the project as completed. The examples in this guide do not provide verified execution evidence from a connected Snowflake account.

Projects 2–5: Customer, JSON, Inventory, and Query Practice

CREATE OR REPLACE TEMP TABLE BA_SF_CUSTOMER_ORDERS AS
SELECT order_id, order_date, customer_id, amount, status
FROM sp_orders_clean
UNION ALL
SELECT '1007', TO_DATE('2026-10-03'), 'C01', 300, 'paid'
UNION ALL
SELECT '1008', TO_DATE('2026-10-04'), 'C02', 200, 'paid';

CREATE OR REPLACE TEMP TABLE BA_SF_CUSTOMER_SUMMARY AS
SELECT customer_id, MIN(order_date) AS first_paid_date,
       MAX(order_date) AS latest_paid_date,
       COUNT(*) AS paid_orders, SUM(amount) AS paid_order_value
FROM BA_SF_CUSTOMER_ORDERS
WHERE status = 'paid'
GROUP BY customer_id;

SELECT *, paid_orders >= 2 AS repeat_buyer
FROM BA_SF_CUSTOMER_SUMMARY ORDER BY customer_id;

SELECT COUNT(*) AS paid_customers,
       COUNT_IF(paid_orders >= 2) AS repeat_buyers,
       ROUND(100.0 * COUNT_IF(paid_orders >= 2)
             / NULLIF(COUNT(*),0),2) AS repeat_buyer_pct
FROM BA_SF_CUSTOMER_SUMMARY;

SELECT IFF(COUNT(*)=3 AND SUM(paid_orders)=5
 AND SUM(paid_order_value)=4500
 AND COALESCE(COUNT_IF(paid_orders>=2),0)=2,'PASS','FAIL') AS customer_baseline
FROM BA_SF_CUSTOMER_SUMMARY;

CREATE OR REPLACE TEMP TABLE BA_SF_TICKET_RAW AS
SELECT PARSE_JSON(column1) AS payload FROM VALUES
 ('{"id":"T1","category":"billing","created":"2026-10-01 09:00:00","resolved":"2026-10-01 11:00:00"}'),
 ('{"id":"T2","category":"billing","created":"2026-10-01 09:00:00","resolved":"2026-10-01 13:00:00"}'),
 ('{"id":"T3","category":"technical","created":"2026-10-01 09:00:00","resolved":"2026-10-01 10:00:00"}'),
 ('{"id":"T4","category":"technical","created":"2026-10-01 09:00:00"}'),
 ('{"id":"T5","category":"billing","created":"2026-10-01 10:00:00","resolved":"2026-10-01 09:00:00"}');

CREATE OR REPLACE TEMP TABLE BA_SF_TICKET_TYPED AS
SELECT payload:id::VARCHAR AS ticket_id,
       payload:category::VARCHAR AS category,
       TRY_TO_TIMESTAMP_NTZ(payload:created::VARCHAR,
         'YYYY-MM-DD HH24:MI:SS') AS created_at,
       payload:resolved::VARCHAR AS resolved_text,
       TRY_TO_TIMESTAMP_NTZ(payload:resolved::VARCHAR,
         'YYYY-MM-DD HH24:MI:SS') AS resolved_at
FROM BA_SF_TICKET_RAW;

CREATE OR REPLACE TEMP TABLE BA_SF_TICKET_RESULTS AS
SELECT *, CASE
  WHEN created_at IS NULL THEN 'invalid'
  WHEN resolved_text IS NULL THEN 'pending'
  WHEN resolved_at IS NULL OR resolved_at < created_at THEN 'invalid'
  ELSE 'closed' END AS record_state,
  DATEDIFF('minute',created_at,resolved_at) AS resolution_minutes
FROM BA_SF_TICKET_TYPED;

SELECT record_state, COUNT(*) AS tickets
FROM BA_SF_TICKET_RESULTS GROUP BY record_state;

SELECT category, COUNT(*) AS closed_tickets,
       MEDIAN(resolution_minutes) AS median_resolution_minutes
FROM BA_SF_TICKET_RESULTS WHERE record_state = 'closed'
GROUP BY category ORDER BY category;

SELECT IFF(COUNT(*)=5
 AND COALESCE(COUNT_IF(record_state='closed'),0)=3
 AND COALESCE(COUNT_IF(record_state='pending'),0)=1
 AND COALESCE(COUNT_IF(record_state='invalid'),0)=1,'PASS','FAIL') AS ticket_baseline
FROM BA_SF_TICKET_RESULTS;

CREATE OR REPLACE TEMP TABLE BA_SF_OPENING (sku VARCHAR, qty INTEGER);
INSERT INTO BA_SF_OPENING VALUES ('A',10),('B',20),('C',5);
CREATE OR REPLACE TEMP TABLE BA_SF_MOVEMENTS
 (source_row INTEGER, movement_id VARCHAR, sku VARCHAR,
  movement_type VARCHAR, qty INTEGER);
INSERT INTO BA_SF_MOVEMENTS VALUES
 (1,'M1','A','receipt',5), (2,'M2','A','shipment',3),
 (3,'M2','A','shipment',3), (4,'M3','B','shipment',4),
 (5,'M4','B','receipt',2), (6,'M5','C','shipment',7);

CREATE OR REPLACE TEMP TABLE BA_SF_MOVEMENTS_CLEAN AS
SELECT * FROM BA_SF_MOVEMENTS
WHERE qty > 0 AND movement_type IN ('receipt','shipment')
QUALIFY ROW_NUMBER() OVER
 (PARTITION BY movement_id ORDER BY source_row DESC) = 1;

CREATE OR REPLACE TEMP TABLE BA_SF_STOCK AS
SELECT o.sku, o.qty AS opening_qty,
       COALESCE(SUM(CASE WHEN m.movement_type = 'receipt'
         THEN m.qty ELSE -m.qty END),0) AS net_movement,
       o.qty + COALESCE(SUM(CASE WHEN m.movement_type = 'receipt'
         THEN m.qty ELSE -m.qty END),0) AS closing_qty
FROM BA_SF_OPENING o LEFT JOIN BA_SF_MOVEMENTS_CLEAN m
  ON o.sku = m.sku
GROUP BY o.sku, o.qty;
SELECT *, closing_qty < 0 AS investigate FROM BA_SF_STOCK ORDER BY sku;

SELECT m.* FROM BA_SF_MOVEMENTS_CLEAN m
LEFT JOIN BA_SF_OPENING o ON m.sku = o.sku
WHERE o.sku IS NULL;
-- Expected: no movements for unknown SKUs

SELECT IFF(COUNT(*)=3
 AND SUM(IFF(sku='A',closing_qty,0))=12
 AND SUM(IFF(sku='B',closing_qty,0))=18
 AND SUM(IFF(sku='C',closing_qty,0))=-2,'PASS','FAIL') AS inventory_baseline
FROM BA_SF_STOCK;

CREATE OR REPLACE TEMP TABLE BA_SF_QUERY_A AS
WITH paid AS (
 SELECT order_date, amount FROM sp_orders_clean
 WHERE status = 'paid'
)
SELECT order_date, SUM(amount) AS paid_order_value
FROM paid GROUP BY order_date;

CREATE OR REPLACE TEMP TABLE BA_SF_QUERY_B AS
SELECT order_date, SUM(amount) AS paid_order_value
FROM sp_orders_clean
WHERE status = 'paid' GROUP BY order_date;

SELECT * FROM BA_SF_QUERY_A MINUS SELECT * FROM BA_SF_QUERY_B;
SELECT * FROM BA_SF_QUERY_B MINUS SELECT * FROM BA_SF_QUERY_A;
-- Both should return zero rows.
SELECT COUNT(*) FROM BA_SF_QUERY_A;
SELECT COUNT(*) FROM BA_SF_QUERY_B;
-- Both should return 2.

SELECT IFF((SELECT COUNT(*) FROM BA_SF_QUERY_A)=2
 AND (SELECT COUNT(*) FROM BA_SF_QUERY_B)=2
 AND (SELECT SUM(paid_order_value) FROM BA_SF_QUERY_A)=4000
 AND (SELECT SUM(paid_order_value) FROM BA_SF_QUERY_B)=4000,
 'PASS','FAIL') AS query_baseline;
-- Retain the two bidirectional MINUS checks as well.

Projects 6–10: Snowflake SQL Projects for Business Analytics

WITH subscriptions AS (
 SELECT column1 AS subscription_id, TO_DATE(column2) AS start_date,
        TO_DATE(column3) AS end_date
 FROM VALUES ('S1','2026-10-01','2026-11-01'),
             ('S2','2026-09-01','2026-10-10'),
             ('S3','2026-10-20','2026-11-20')
), params AS (SELECT TO_DATE('2026-10-15') AS as_of_date)
SELECT subscription_id, start_date, end_date,
       end_date <= DATEADD('day',30,as_of_date) AS expires_within_30_days
FROM subscriptions CROSS JOIN params
WHERE start_date <= as_of_date AND end_date > as_of_date;
-- S1 only; expires_within_30_days = TRUE

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P06_RESULT AS
WITH subscriptions AS (
 SELECT column1 AS subscription_id, TO_DATE(column2) AS start_date,
        TO_DATE(column3) AS end_date
 FROM VALUES ('S1','2026-10-01','2026-11-01'),
             ('S2','2026-09-01','2026-10-10'),
             ('S3','2026-10-20','2026-11-20')
), params AS (SELECT TO_DATE('2026-10-15') AS as_of_date)
SELECT subscription_id, start_date, end_date,
       end_date <= DATEADD('day',30,as_of_date) AS expires_within_30_days
FROM subscriptions CROSS JOIN params
WHERE start_date <= as_of_date AND end_date > as_of_date;

SELECT IFF(COUNT(*)=1 AND MIN(subscription_id)='S1'
 AND COALESCE(COUNT_IF(expires_within_30_days),0)=1,'PASS','FAIL') AS project_6_baseline
FROM BA_SF_P06_RESULT;

WITH invoices AS (
 SELECT column1 AS invoice_id, column2::NUMBER(12,2) AS invoice_amount
 FROM VALUES ('I1',1000),('I2',500)
), payments AS (
 SELECT column1 AS payment_id, column2 AS invoice_id,
        column3::NUMBER(12,2) AS amount
 FROM VALUES ('P1','I1',400),('P2','I1',300),('P3','I2',600)
), totals AS (
 SELECT invoice_id, SUM(amount) AS paid FROM payments GROUP BY invoice_id
)
SELECT i.invoice_id, i.invoice_amount, COALESCE(t.paid,0) AS paid,
       i.invoice_amount - COALESCE(t.paid,0) AS balance_due
FROM invoices i LEFT JOIN totals t ON i.invoice_id = t.invoice_id
ORDER BY i.invoice_id;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P07_RESULT AS
WITH invoices AS (
 SELECT column1 AS invoice_id, column2::NUMBER(12,2) AS invoice_amount
 FROM VALUES ('I1',1000),('I2',500)
), payments AS (
 SELECT column1 AS payment_id, column2 AS invoice_id,
        column3::NUMBER(12,2) AS amount
 FROM VALUES ('P1','I1',400),('P2','I1',300),('P3','I2',600)
), totals AS (
 SELECT invoice_id, SUM(amount) AS paid FROM payments GROUP BY invoice_id
)
SELECT i.invoice_id, i.invoice_amount, COALESCE(t.paid,0) AS paid,
       i.invoice_amount - COALESCE(t.paid,0) AS balance_due
FROM invoices i LEFT JOIN totals t ON i.invoice_id = t.invoice_id
ORDER BY i.invoice_id;

SELECT IFF(COUNT(*)=2 AND SUM(invoice_amount)=1500
 AND SUM(paid)=1300 AND SUM(balance_due)=200
 AND MIN(IFF(invoice_id='I1',balance_due,NULL))=300
 AND MIN(IFF(invoice_id='I2',balance_due,NULL))=-100,'PASS','FAIL') AS project_7_baseline
FROM BA_SF_P07_RESULT;

WITH shipments AS (
 SELECT column1 AS shipment_id,
        TO_TIMESTAMP_NTZ(column2) AS promised_at,
        TO_TIMESTAMP_NTZ(column3) AS delivered_at
 FROM VALUES
 ('D1','2026-10-01 12:00:00','2026-10-01 11:30:00'),
 ('D2','2026-10-01 12:00:00','2026-10-01 12:20:00'),
 ('D3','2026-10-01 12:00:00',NULL)
)
SELECT COUNT_IF(delivered_at IS NOT NULL) AS completed,
       COUNT_IF(delivered_at <= promised_at) AS on_time,
       ROUND(100.0 * COUNT_IF(delivered_at <= promised_at)
        / NULLIF(COUNT_IF(delivered_at IS NOT NULL),0),2) AS on_time_pct
FROM shipments;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P08_RESULT AS
WITH shipments AS (
 SELECT column1 AS shipment_id,
        TO_TIMESTAMP_NTZ(column2) AS promised_at,
        TO_TIMESTAMP_NTZ(column3) AS delivered_at
 FROM VALUES
 ('D1','2026-10-01 12:00:00','2026-10-01 11:30:00'),
 ('D2','2026-10-01 12:00:00','2026-10-01 12:20:00'),
 ('D3','2026-10-01 12:00:00',NULL)
)
SELECT COUNT_IF(delivered_at IS NOT NULL) AS completed,
       COUNT_IF(delivered_at <= promised_at) AS on_time,
       ROUND(100.0 * COUNT_IF(delivered_at <= promised_at)
        / NULLIF(COUNT_IF(delivered_at IS NOT NULL),0),2) AS on_time_pct
FROM shipments;

SELECT IFF(COUNT(*)=1 AND MIN(completed)=2 AND MIN(on_time)=1
 AND MIN(on_time_pct)=50,'PASS','FAIL') AS project_8_baseline
FROM BA_SF_P08_RESULT;

WITH events AS (
 SELECT column1 AS session_id, column2 AS event_name,
        TO_TIMESTAMP_NTZ(column3) AS event_at FROM VALUES
 ('A','view','2026-10-01 10:00:00'),
 ('A','cart','2026-10-01 10:01:00'),
 ('A','purchase','2026-10-01 10:03:00'),
 ('B','view','2026-10-01 11:00:00'),
 ('C','cart','2026-10-01 12:00:00'),
 ('C','purchase','2026-10-01 12:01:00')
), stages AS (
 SELECT session_id,
  MIN(IFF(event_name='view',event_at,NULL)) AS viewed_at,
  MIN(IFF(event_name='cart',event_at,NULL)) AS cart_at,
  MIN(IFF(event_name='purchase',event_at,NULL)) AS purchased_at
 FROM events GROUP BY session_id
)
SELECT COUNT_IF(viewed_at IS NOT NULL) AS viewed_sessions,
       COUNT_IF(viewed_at <= cart_at AND cart_at <= purchased_at)
         AS complete_funnels
FROM stages;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P09_RESULT AS
WITH events AS (
 SELECT column1 AS session_id, column2 AS event_name,
        TO_TIMESTAMP_NTZ(column3) AS event_at FROM VALUES
 ('A','view','2026-10-01 10:00:00'),
 ('A','cart','2026-10-01 10:01:00'),
 ('A','purchase','2026-10-01 10:03:00'),
 ('B','view','2026-10-01 11:00:00'),
 ('C','cart','2026-10-01 12:00:00'),
 ('C','purchase','2026-10-01 12:01:00')
), stages AS (
 SELECT session_id,
  MIN(IFF(event_name='view',event_at,NULL)) AS viewed_at,
  MIN(IFF(event_name='cart',event_at,NULL)) AS cart_at,
  MIN(IFF(event_name='purchase',event_at,NULL)) AS purchased_at
 FROM events GROUP BY session_id
)
SELECT COUNT_IF(viewed_at IS NOT NULL) AS viewed_sessions,
       COUNT_IF(viewed_at <= cart_at AND cart_at <= purchased_at)
         AS complete_funnels
FROM stages;

SELECT IFF(COUNT(*)=1 AND MIN(viewed_sessions)=2 AND MIN(complete_funnels)=1,'PASS','FAIL') AS project_9_baseline
FROM BA_SF_P09_RESULT;

WITH spend AS (
 SELECT column1 AS campaign, column2 AS spend_amount
 FROM VALUES ('A',100),('B',200),('C',50)
), orders AS (
 SELECT column1 AS order_id, column2 AS campaign, column3 AS amount
 FROM VALUES ('O1','A',150),('O2','A',50),('O3','B',100)
), attributed AS (
 SELECT campaign, SUM(amount) AS order_value FROM orders GROUP BY campaign
)
SELECT s.campaign, s.spend_amount,
       COALESCE(a.order_value,0) AS attributed_order_value,
       ROUND(COALESCE(a.order_value,0)/NULLIF(s.spend_amount,0),2)
         AS attributed_value_per_spend
FROM spend s LEFT JOIN attributed a ON s.campaign=a.campaign
ORDER BY s.campaign;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P10_RESULT AS
WITH spend AS (
 SELECT column1 AS campaign, column2 AS spend_amount
 FROM VALUES ('A',100),('B',200),('C',50)
), orders AS (
 SELECT column1 AS order_id, column2 AS campaign, column3 AS amount
 FROM VALUES ('O1','A',150),('O2','A',50),('O3','B',100)
), attributed AS (
 SELECT campaign, SUM(amount) AS order_value FROM orders GROUP BY campaign
)
SELECT s.campaign, s.spend_amount,
       COALESCE(a.order_value,0) AS attributed_order_value,
       ROUND(COALESCE(a.order_value,0)/NULLIF(s.spend_amount,0),2)
         AS attributed_value_per_spend
FROM spend s LEFT JOIN attributed a ON s.campaign=a.campaign
ORDER BY s.campaign;

SELECT IFF(COUNT(*)=3 AND SUM(spend_amount)=350
 AND SUM(attributed_order_value)=300
 AND MIN(IFF(campaign='A',attributed_value_per_spend,NULL))=2
 AND MIN(IFF(campaign='B',attributed_value_per_spend,NULL))=0.5,'PASS','FAIL') AS project_10_baseline
FROM BA_SF_P10_RESULT;

Projects 11–15: Operations and Event Analytics

WITH shipped AS (
 SELECT column1 AS unit_id FROM VALUES ('U1'),('U2'),('U3')
), returns AS (
 SELECT column1 AS event_id, column2 AS unit_id, column3 AS reason
 FROM VALUES ('R1','U1','damaged'),('R1','U1','damaged'),
             ('R2','U2','wrong_size')
), returned_units AS (SELECT DISTINCT unit_id FROM returns)
SELECT COUNT(*) AS shipped_units,
       COUNT(r.unit_id) AS returned_units,
       ROUND(100.0*COUNT(r.unit_id)/NULLIF(COUNT(*),0),2) AS return_pct
FROM shipped s LEFT JOIN returned_units r ON s.unit_id=r.unit_id;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P11_RESULT AS
WITH shipped AS (
 SELECT column1 AS unit_id FROM VALUES ('U1'),('U2'),('U3')
), returns AS (
 SELECT column1 AS event_id, column2 AS unit_id, column3 AS reason
 FROM VALUES ('R1','U1','damaged'),('R1','U1','damaged'),
             ('R2','U2','wrong_size')
), returned_units AS (SELECT DISTINCT unit_id FROM returns)
SELECT COUNT(*) AS shipped_units,
       COUNT(r.unit_id) AS returned_units,
       ROUND(100.0*COUNT(r.unit_id)/NULLIF(COUNT(*),0),2) AS return_pct
FROM shipped s LEFT JOIN returned_units r ON s.unit_id=r.unit_id;

SELECT IFF(COUNT(*)=1 AND MIN(shipped_units)=3 AND MIN(returned_units)=2
 AND MIN(return_pct)=66.67,'PASS','FAIL') AS project_11_baseline
FROM BA_SF_P11_RESULT;

WITH ordered AS (
 SELECT column1 AS po_id, column2 AS line_id, column3 AS ordered_qty
 FROM VALUES ('PO1','A',10),('PO1','B',5)
), receipts AS (
 SELECT column1 AS receipt_id, column2 AS po_id,
        column3 AS line_id, column4 AS qty
 FROM VALUES ('R1','PO1','A',4),('R2','PO1','A',3),('R3','PO1','B',6)
), received AS (
 SELECT po_id,line_id,SUM(qty) AS received_qty
 FROM receipts GROUP BY po_id,line_id
)
SELECT o.po_id,o.line_id,o.ordered_qty,COALESCE(r.received_qty,0) AS received_qty,
       o.ordered_qty-COALESCE(r.received_qty,0) AS remaining_qty
FROM ordered o LEFT JOIN received r
 ON o.po_id=r.po_id AND o.line_id=r.line_id ORDER BY o.line_id;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P12_RESULT AS
WITH ordered AS (
 SELECT column1 AS po_id, column2 AS line_id, column3 AS ordered_qty
 FROM VALUES ('PO1','A',10),('PO1','B',5)
), receipts AS (
 SELECT column1 AS receipt_id, column2 AS po_id,
        column3 AS line_id, column4 AS qty
 FROM VALUES ('R1','PO1','A',4),('R2','PO1','A',3),('R3','PO1','B',6)
), received AS (
 SELECT po_id,line_id,SUM(qty) AS received_qty
 FROM receipts GROUP BY po_id,line_id
)
SELECT o.po_id,o.line_id,o.ordered_qty,COALESCE(r.received_qty,0) AS received_qty,
       o.ordered_qty-COALESCE(r.received_qty,0) AS remaining_qty
FROM ordered o LEFT JOIN received r
 ON o.po_id=r.po_id AND o.line_id=r.line_id ORDER BY o.line_id;

SELECT IFF(COUNT(*)=2 AND SUM(ordered_qty)=15 AND SUM(received_qty)=13
 AND MIN(IFF(line_id='A',remaining_qty,NULL))=3
 AND MIN(IFF(line_id='B',remaining_qty,NULL))=-1,'PASS','FAIL') AS project_12_baseline
FROM BA_SF_P12_RESULT;

WITH enrolled AS (
 SELECT column1 AS learner_id FROM VALUES ('L1'),('L2'),('L3')
), attendance AS (
 SELECT column1 AS learner_id,column2 AS session_id
 FROM VALUES ('L1','S1'),('L1','S1'),('L1','S2'),('L2','S1')
), unique_attendance AS (SELECT DISTINCT learner_id,session_id FROM attendance),
 totals AS (
 SELECT learner_id,COUNT(*) AS attended FROM unique_attendance GROUP BY learner_id
)
SELECT e.learner_id,COALESCE(t.attended,0) AS attended,
       ROUND(100.0*COALESCE(t.attended,0)/2,2) AS attendance_pct
FROM enrolled e LEFT JOIN totals t ON e.learner_id=t.learner_id
ORDER BY e.learner_id;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P13_RESULT AS
WITH enrolled AS (
 SELECT column1 AS learner_id FROM VALUES ('L1'),('L2'),('L3')
), attendance AS (
 SELECT column1 AS learner_id,column2 AS session_id
 FROM VALUES ('L1','S1'),('L1','S1'),('L1','S2'),('L2','S1')
), unique_attendance AS (SELECT DISTINCT learner_id,session_id FROM attendance),
 totals AS (
 SELECT learner_id,COUNT(*) AS attended FROM unique_attendance GROUP BY learner_id
)
SELECT e.learner_id,COALESCE(t.attended,0) AS attended,
       ROUND(100.0*COALESCE(t.attended,0)/2,2) AS attendance_pct
FROM enrolled e LEFT JOIN totals t ON e.learner_id=t.learner_id
ORDER BY e.learner_id;

SELECT IFF(COUNT(*)=3 AND SUM(attended)=3
 AND MIN(IFF(learner_id='L1',attendance_pct,NULL))=100
 AND MIN(IFF(learner_id='L2',attendance_pct,NULL))=50
 AND MIN(IFF(learner_id='L3',attendance_pct,NULL))=0,'PASS','FAIL') AS project_13_baseline
FROM BA_SF_P13_RESULT;

WITH requests AS (
 SELECT column1 AS request_id,column2 AS endpoint,
        column3 AS status_code,column4 AS latency_ms
 FROM VALUES ('Q1','/orders',200,100),('Q2','/orders',500,900),
             ('Q3','/orders',200,200),('Q4','/health',200,20)
)
SELECT endpoint,COUNT(*) AS requests,
       ROUND(100.0*COALESCE(COUNT_IF(status_code >= 500 AND status_code < 600),0)
          /COUNT(*),2) AS server_error_pct,
       MEDIAN(latency_ms) AS median_ms,
       PERCENTILE_CONT(0.95) WITHIN GROUP (ORDER BY latency_ms) AS p95_ms
FROM requests GROUP BY endpoint ORDER BY endpoint;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P14_RESULT AS
WITH requests AS (
 SELECT column1 AS request_id,column2 AS endpoint,
        column3 AS status_code,column4 AS latency_ms
 FROM VALUES ('Q1','/orders',200,100),('Q2','/orders',500,900),
             ('Q3','/orders',200,200),('Q4','/health',200,20)
)
SELECT endpoint,COUNT(*) AS requests,
       ROUND(100.0*COALESCE(COUNT_IF(status_code >= 500 AND status_code < 600),0)
          /COUNT(*),2) AS server_error_pct,
       MEDIAN(latency_ms) AS median_ms,
       PERCENTILE_CONT(0.95) WITHIN GROUP (ORDER BY latency_ms) AS p95_ms
FROM requests GROUP BY endpoint ORDER BY endpoint;

SELECT IFF(COUNT(*)=2
 AND MIN(IFF(endpoint='/orders',requests,NULL))=3
 AND ABS(MIN(IFF(endpoint='/orders',p95_ms,NULL))-830)<0.0001
 AND MIN(IFF(endpoint='/health',server_error_pct,NULL))=0,'PASS','FAIL') AS project_14_baseline
FROM BA_SF_P14_RESULT;

WITH readings AS (
 SELECT column1 AS sensor_id,TO_TIMESTAMP_NTZ(column2) AS observed_at,
        column3 AS reading FROM VALUES
 ('S1','2026-10-01 10:00:00',20),
 ('S1','2026-10-01 10:05:00',85),
 ('S1','2026-10-01 10:20:00',25)
), gaps AS (
 SELECT *,DATEDIFF('minute',LAG(observed_at) OVER
   (PARTITION BY sensor_id ORDER BY observed_at),observed_at) AS gap_minutes
 FROM readings
)
SELECT *,reading > 80 AS threshold_breach,
       gap_minutes > 5 AS missing_interval FROM gaps ORDER BY observed_at;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P15_RESULT AS
WITH readings AS (
 SELECT column1 AS sensor_id,TO_TIMESTAMP_NTZ(column2) AS observed_at,
        column3 AS reading FROM VALUES
 ('S1','2026-10-01 10:00:00',20),
 ('S1','2026-10-01 10:05:00',85),
 ('S1','2026-10-01 10:20:00',25)
), gaps AS (
 SELECT *,DATEDIFF('minute',LAG(observed_at) OVER
   (PARTITION BY sensor_id ORDER BY observed_at),observed_at) AS gap_minutes
 FROM readings
)
SELECT *,reading > 80 AS threshold_breach,
       gap_minutes > 5 AS missing_interval FROM gaps ORDER BY observed_at;

SELECT IFF(COUNT(*)=3 AND COALESCE(COUNT_IF(threshold_breach),0)=1
 AND COALESCE(COUNT_IF(missing_interval),0)=1
 AND MAX(gap_minutes)=15,'PASS','FAIL') AS project_15_baseline
FROM BA_SF_P15_RESULT;

Projects 16–20: Snowflake Data Engineering Projects

CREATE OR REPLACE TEMP TABLE BA_SF_CUSTOMER_CURRENT
 (customer_id VARCHAR,city VARCHAR,updated_at TIMESTAMP_NTZ);
INSERT INTO BA_SF_CUSTOMER_CURRENT VALUES
 ('C1','Hyderabad','2026-10-01 09:00:00');
CREATE OR REPLACE TEMP TABLE BA_SF_CUSTOMER_FEED
 (source_seq INTEGER,customer_id VARCHAR,city VARCHAR,updated_at TIMESTAMP_NTZ);
INSERT INTO BA_SF_CUSTOMER_FEED VALUES
 (1,'C1','Mumbai','2026-10-02 09:00:00'),
 (2,'C1','Pune','2026-10-03 09:00:00'),
 (3,'C2','Chennai','2026-10-02 09:00:00');

MERGE INTO BA_SF_CUSTOMER_CURRENT t USING (
 SELECT * FROM BA_SF_CUSTOMER_FEED
 QUALIFY ROW_NUMBER() OVER (PARTITION BY customer_id
  ORDER BY updated_at DESC,source_seq DESC)=1
) s ON t.customer_id=s.customer_id
WHEN MATCHED AND s.updated_at > t.updated_at THEN
 UPDATE SET city=s.city,updated_at=s.updated_at
WHEN NOT MATCHED THEN
 INSERT (customer_id,city,updated_at) VALUES (s.customer_id,s.city,s.updated_at);
SELECT * FROM BA_SF_CUSTOMER_CURRENT ORDER BY customer_id;

SELECT IFF(COUNT(*)=2 AND COUNT(DISTINCT customer_id)=2
 AND MIN(IFF(customer_id='C1',city,NULL))='Pune'
 AND MIN(IFF(customer_id='C2',city,NULL))='Chennai',
 'PASS','FAIL') AS current_customer_baseline
FROM BA_SF_CUSTOMER_CURRENT;

WITH history AS (
 SELECT column1 AS customer_id,column2 AS city,
        TO_DATE(column3) AS valid_from,TO_DATE(column4) AS valid_to
 FROM VALUES ('C1','Hyderabad','2026-10-01','2026-10-10'),
             ('C1','Pune','2026-10-10',NULL)
), orders AS (
 SELECT column1 AS order_id,column2 AS customer_id,TO_DATE(column3) AS order_date
 FROM VALUES ('O1','C1','2026-10-05'),('O2','C1','2026-10-12')
)
SELECT o.order_id,o.order_date,h.city
FROM orders o LEFT JOIN history h
 ON o.customer_id=h.customer_id AND o.order_date >= h.valid_from
 AND (o.order_date < h.valid_to OR h.valid_to IS NULL)
ORDER BY o.order_id;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P17_RESULT AS
WITH history AS (
 SELECT column1 AS customer_id,column2 AS city,
        TO_DATE(column3) AS valid_from,TO_DATE(column4) AS valid_to
 FROM VALUES ('C1','Hyderabad','2026-10-01','2026-10-10'),
             ('C1','Pune','2026-10-10',NULL)
), orders AS (
 SELECT column1 AS order_id,column2 AS customer_id,TO_DATE(column3) AS order_date
 FROM VALUES ('O1','C1','2026-10-05'),('O2','C1','2026-10-12')
)
SELECT o.order_id,o.order_date,h.city
FROM orders o LEFT JOIN history h
 ON o.customer_id=h.customer_id AND o.order_date >= h.valid_from
 AND (o.order_date < h.valid_to OR h.valid_to IS NULL)
ORDER BY o.order_id;

SELECT IFF(COUNT(*)=2 AND COUNT(DISTINCT order_id)=2
 AND MIN(IFF(order_id='O1',city,NULL))='Hyderabad'
 AND MIN(IFF(order_id='O2',city,NULL))='Pune','PASS','FAIL') AS project_17_baseline
FROM BA_SF_P17_RESULT;

WITH source_control AS (
 SELECT column1 AS batch_id,column2 AS row_count,column3 AS amount
 FROM VALUES ('B1',3,300),('B2',2,200)
), target_control AS (
 SELECT column1 AS batch_id,column2 AS row_count,column3 AS amount
 FROM VALUES ('B1',3,300),('B2',1,100),('B3',1,50)
)
SELECT COALESCE(s.batch_id,t.batch_id) AS batch_id,
       s.row_count AS source_rows,t.row_count AS target_rows,
       t.amount-s.amount AS value_difference,
       CASE WHEN s.batch_id IS NULL THEN 'unexpected target'
            WHEN t.batch_id IS NULL THEN 'missing target'
            WHEN s.row_count=t.row_count AND s.amount=t.amount THEN 'match'
            ELSE 'mismatch' END AS audit_state
FROM source_control s FULL OUTER JOIN target_control t
 ON s.batch_id=t.batch_id ORDER BY batch_id;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P18_RESULT AS
WITH source_control AS (
 SELECT column1 AS batch_id,column2 AS row_count,column3 AS amount
 FROM VALUES ('B1',3,300),('B2',2,200)
), target_control AS (
 SELECT column1 AS batch_id,column2 AS row_count,column3 AS amount
 FROM VALUES ('B1',3,300),('B2',1,100),('B3',1,50)
)
SELECT COALESCE(s.batch_id,t.batch_id) AS batch_id,
       s.row_count AS source_rows,t.row_count AS target_rows,
       t.amount-s.amount AS value_difference,
       CASE WHEN s.batch_id IS NULL THEN 'unexpected target'
            WHEN t.batch_id IS NULL THEN 'missing target'
            WHEN s.row_count=t.row_count AND s.amount=t.amount THEN 'match'
            ELSE 'mismatch' END AS audit_state
FROM source_control s FULL OUTER JOIN target_control t
 ON s.batch_id=t.batch_id ORDER BY batch_id;

SELECT IFF(COUNT(*)=3
 AND COALESCE(COUNT_IF(audit_state='match'),0)=1
 AND COALESCE(COUNT_IF(audit_state='mismatch'),0)=1
 AND COALESCE(COUNT_IF(audit_state='unexpected target'),0)=1
 AND MIN(IFF(batch_id='B2',value_difference,NULL))=-100,'PASS','FAIL') AS project_18_baseline
FROM BA_SF_P18_RESULT;

WITH orders AS (
 SELECT column1 AS order_id,column2 AS currency,
        column3::NUMBER(12,2) AS amount,TO_DATE('2026-10-01') AS rate_date
 FROM VALUES ('O1','USD',100),('O2','EUR',100),('O3','GBP',100)
), rates AS (
 SELECT column1 AS currency,column2::NUMBER(12,6) AS usd_per_unit,
        TO_DATE('2026-10-01') AS rate_date
 FROM VALUES ('USD',1.0),('EUR',1.1)
)
SELECT o.order_id,o.currency,o.amount,r.usd_per_unit,
       ROUND(o.amount*r.usd_per_unit,2) AS usd_amount,
       IFF(r.currency IS NULL,'missing rate','converted') AS conversion_state
FROM orders o LEFT JOIN rates r
 ON o.currency=r.currency AND o.rate_date=r.rate_date
ORDER BY o.order_id;

-- Keep the original seed unchanged for this baseline.
-- Persist the starter result in the SAME session:
CREATE OR REPLACE TEMP TABLE BA_SF_P19_RESULT AS
WITH orders AS (
 SELECT column1 AS order_id,column2 AS currency,
        column3::NUMBER(12,2) AS amount,TO_DATE('2026-10-01') AS rate_date
 FROM VALUES ('O1','USD',100),('O2','EUR',100),('O3','GBP',100)
), rates AS (
 SELECT column1 AS currency,column2::NUMBER(12,6) AS usd_per_unit,
        TO_DATE('2026-10-01') AS rate_date
 FROM VALUES ('USD',1.0),('EUR',1.1)
)
SELECT o.order_id,o.currency,o.amount,r.usd_per_unit,
       ROUND(o.amount*r.usd_per_unit,2) AS usd_amount,
       IFF(r.currency IS NULL,'missing rate','converted') AS conversion_state
FROM orders o LEFT JOIN rates r
 ON o.currency=r.currency AND o.rate_date=r.rate_date
ORDER BY o.order_id;

SELECT IFF(COUNT(*)=3
 AND COALESCE(COUNT_IF(conversion_state='converted'),0)=2
 AND COALESCE(COUNT_IF(conversion_state='missing rate'),0)=1
 AND SUM(usd_amount)=210,'PASS','FAIL') AS project_19_baseline
FROM BA_SF_P19_RESULT;

CREATE OR REPLACE TEMP VIEW BA_SF_DAILY_PAID_REPORT AS
SELECT order_date AS paid_order_date,
       COUNT(*) AS paid_order_count,
       SUM(amount)::NUMBER(12,2) AS paid_order_value
FROM sp_orders_clean
WHERE status='paid'
GROUP BY order_date;

SELECT * FROM BA_SF_DAILY_PAID_REPORT ORDER BY paid_order_date;
SELECT SUM(paid_order_count) AS orders,SUM(paid_order_value) AS value
FROM BA_SF_DAILY_PAID_REPORT;

SELECT IFF(COUNT(*)=2 AND COUNT(DISTINCT paid_order_date)=2
 AND SUM(paid_order_count)=3 AND SUM(paid_order_value)=4000,
 'PASS','FAIL') AS reporting_view_baseline
FROM BA_SF_DAILY_PAID_REPORT;

Advanced Snowflake Project Extensions

These capstones are implementation briefs. They define scope and acceptance criteria rather than providing complete deployed pipelines.

Capstone A: Monitored Order-to-Report Pipeline

Extend Project 1 with file ingestion, incremental order processing, batch controls, and reporting.

Define event keys, version rules, and correction handling before selecting orchestration.

Demonstrate that:

• Reprocessing a batch does not inflate totals.
• Interrupted processing can resume with an audit trail.
• Newer valid corrections update the intended records.
• Older records do not overwrite newer accepted state.
• Missing and failed batches appear in monitoring.

Retain architecture, scripts, run logs, reconciliation, recovery evidence, and actual usage observations.

Capstone B: Historical Customer and Multi-Currency Reporting

Combine historical address reporting with currency conversion.

Keep order, history, and rate keys separate. Each order should match at most one valid address and one approved rate.

Test boundary timestamps, overlapping history, duplicate rates, and missing reference data. Show coverage measures beside converted results.

Retain the data model, acceptance queries, exceptions, correction policy, and consumer report.

Troubleshoot Common Project Problems

Change one condition at a time. Record the expected behaviour, actual result, and correction.

Build Snowflake Portfolio Projects You Can Explain

A useful portfolio lets another person reproduce your work and understand why the result is trustworthy.

Organize Your Repository

snowflake-project-portfolio/
├── README.md
├── data/
├── sql/
│   ├── 01_setup_and_load.sql
│   ├── 02_transform.sql
│   ├── 03_validate.sql
│   └── 04_report.sql
├── docs/
│   ├── architecture.md
│   ├── data-dictionary.csv
│   └── decisions.md
└── evidence/
    ├── baseline-results.csv
    └── test-log.csv

Document the business problem, prerequisites, dataset source, run order, keys, row grain, rules, expected outputs, and limitations.

Identify scripts that reset objects. Exclude credentials and private records.

Include Actual Result Evidence

Capture input counts, rejected records, final reports, reconciliation checks, and at least one changed-input test.

Add screenshots or exports from your completed implementation. Explain what each item proves.

Present Snowflake Projects on Your Résumé

Describe the work you completed and the checks you performed.

Sales-data quality example:

Built a Snowflake SQL portfolio project that validated synthetic order data, retained rejected records, removed duplicate orders, and reconciled daily paid-order totals.

Customer loading example:

Implemented customer-feed loading with deterministic source selection and tests for unchanged reruns and older updates.

Use results from your own work. Avoid unsupported claims about production scale, client experience, or performance gains.

Prepare for Project Interviews

Be ready to explain the business question, table grain, duplicate policy, validation approach, and failure handling.

Discuss what you implemented independently and what would still be required for production use. For further preparation, explore Brolly Academy’s Snowflake Interview Questions.

Continue Learning With Brolly Academy

Use these projects to identify your next learning goal: SQL foundations, semi-structured data, business reporting, or data engineering.

Explore Brolly Academy’s Snowflake Training in Hyderabad for structured learning.

Share your current SQL level and the project you want to complete. Request the current project list, lab arrangements, review process, and support schedule for your selected batch.

Aim for a working project, validated results, clear documentation, and the ability to explain your decisions.

Frequently Asked Questions

Which Snowflake project is best for beginners?

Start with sales-data quality. It combines loading, conversion, validation, deduplication, and reporting in one manageable exercise.

Can I build Snowflake projects with SQL alone?

Yes. The first walkthrough uses SQL to load synthetic records, transform them, retain rejected rows, and produce a report. External tools are optional extensions.

Where can I find Snowflake projects with datasets?

The tutorials include synthetic sample inputs in their SQL examples. Projects 2, 5 and 20 reuse the clean sales dataset from Project 1. For an extension, use synthetic or permitted public data and document its source.

Does this guide include source code for every project?

All 20 projects include SQL examples on this page, with sample inputs, expected results, and checks. Projects 2, 5, and 20 reuse the clean orders from Project 1. Save the scripts in your own repository and execute the checks in your practice account.

Have the SQL examples been executed in Snowflake?

This guide does not provide verified execution evidence from a connected Snowflake account. Run the examples in your workspace and retain actual outputs before presenting them as completed work.

Are these projects free?

The article’s examples are available on this page. Snowflake compute, storage, and optional services may incur charges depending on your account.

What should a Snowflake end-to-end project include?

Include source data, loading, transformations, validation, reporting, and reproduction instructions. Operational extensions also need scheduling, monitoring, access controls, and recovery.

How many projects should I include in my portfolio?

Start with one complete project. Add another when it demonstrates a different skill, such as JSON processing, historical joins, or incremental loading.

Do these projects require Python or Power BI?

The SQL walkthrough does not require either. Add Python for a defined processing need or BI software for a reporting requirement.

Are these production-ready pipelines?

They are learning exercises and implementation briefs. Production readiness requires controls appropriate to the source systems, users, workload, and operating environment.

Does Brolly Academy provide project guidance?

Confirm the project guidance, reviews, lab access, and support included in your selected batch through the Snowflake training page.

Continue with our detailed guides to dynamic tables in Snowflake, Snowflake Cortex AI, Snowflake streams and tasks and Snowflake Time Travel for focused SQL walkthroughs, expected outputs and validation exercises.

Technical References

Use official Snowflake documentation to check syntax and platform behaviour:

• Snowflake Documentation
• Temporary and Transient Tables
• TRY_TO_DATE
• ROW_NUMBER
• QUALIFY
• MERGE

Related Snowflake Articles

Continue with the Snowpipe S3 auto-ingest walkthrough and the Snowflake query optimization SQL lab.

Start with the Snowflake prerequisites and SQL readiness guide, then follow the step-by-step guide to learning Snowflake before choosing a larger project.

Related Courses