FreeInteractive

SQL for AI/ML Production

Name: SQL for AI/ML Production
Availability: InStock

with LaunchFar AI

5 weeks46 lessons

Learn the SQL you'll actually run on the job — querying an AI startup's warehouse: LLM cost ledgers, prompt-template versions, retrieval traces, eval runs, retention cohorts. Every lesson is one challenge against the same warehouse you'd see in production. Personalized to your skill level so you only see the modules you actually need.

Syllabus

1. Meet the warehouse
- Meet the warehouse6 min
2. First queries
- Pull recent LLM calls5 min
- Debug: the spot-check slice comes back wrong4 min
- Distinct models this week4 min
- Top 20 most expensive calls5 min
- Page 2 of the leaderboard4 min
3. Filtering down to the rows you actually want
- Only haiku-4.5 calls4 min
- Expensive calls only4 min
- Cheap and fast5 min
- Top-tier models only4 min
- Last week of April5 min
4. Missing values, text & basic aggregation
- Find the unrated calls4 min
- Default rating to zero4 min
- Search by template name4 min
- Debug: the daily call count reads zero4 min
- Cost stats for the month5 min
5. Spend per model
- Spend per model8 min
- Cost and latency per model6 min
- Only models over budget6 min
- Spend per model per day7 min
- WHERE vs HAVING7 min
6. Joining tables for the full picture
- Attach the template name6 min
- Templates with zero calls6 min
- User → session → event walk7 min
- Compare versions of the same template6 min
- Spend per template (with name)7 min
7. Subqueries and conditional logic
- Above-average-cost calls6 min
- Percent-of-overall next to each row6 min
- Users who ever rated5 min
- Bucket cost into tiers5 min
- Combine prompt calls and events on one timeline6 min
8. Time, recency, and basic windows
- Truncate to hour and day5 min
- Latest call per user7 min
- Rank templates by spend (RANK vs DENSE_RANK)7 min
- Daily active users6 min
- Cohort month vs active month7 min
9. Running total cost by day
- Running total cost by day10 min
- Compare this call to the last7 min
- Moving average latency, trailing 7 rows8 min
- Percent of total spend by model6 min
10. CTEs, recursion, and query optimization
- Name the steps with a CTE chain7 min
- Walk prompt template versions (recursive CTE)8 min
- Read the query plan and add an index8 min
11. Capstones and DML for analysts
- LLM cost leaderboard with week-over-week9 min
- Capstone: the April template report15 min
- INSERT, UPDATE, DELETE for analysts8 min

Start this course

LaunchFar is invite-only during the beta. Join the waitlist to get your spot — the first 250 members get our first course free.

Join the waitlist

1. Meet the warehouse

2. First queries

3. Filtering down to the rows you actually want

4. Missing values, text & basic aggregation

5. Spend per model

6. Joining tables for the full picture

7. Subqueries and conditional logic

8. Time, recency, and basic windows

9. Running total cost by day

10. CTEs, recursion, and query optimization

11. Capstones and DML for analysts

Start this course