Overview
Who it is for: anyone who knows a little Python and wants to understand, rebuild or extend a production-style AI application, from beginners to working engineers. Each module says what to read first.
What you will learn
- 20 written parts, yours for good
- 6h 59m of reading, measured not estimated
- Written for the intermediate level
- Every future revision included
Curriculum
20 sections · 20 parts · 6h 59m29 min free to read · 6h 30m with the project
01Start here1 part · 2 min
02Project overview: the Kestrel Bank AI assistant1 part · 7 min
- Project overview: the Kestrel Bank AI assistantPreview7 min
03Module 1: The problem a bank assistant solves1 part · 20 min
- Module 1: The problem a bank assistant solvesPreview20 min
04Module 2: The solution: retrieval, tools and approval1 part · 21 min
- Module 2: The solution: retrieval, tools and approval21 min
05Module 3: Infrastructure decisions and their trade-offs1 part · 22 min
- Module 3: Infrastructure decisions and their trade-offs22 min
06Module 4: High-level design: components and flows1 part · 23 min
- Module 4: High-level design: components and flows23 min
07Module 5: Low-level design: data, API and state1 part · 22 min
- Module 5: Low-level design: data, API and state22 min
08Module 6: Authentication and role-based access control1 part · 24 min
- Module 6: Authentication and role-based access control24 min
09Module 7: Core banking: the ledger, transfers and synthetic customers1 part · 25 min
- Module 7: Core banking: the ledger, transfers and synthetic customers25 min
10Module 8: Document ingestion, from file to chunks1 part · 24 min
- Module 8: Document ingestion, from file to chunks24 min
11Module 9: Hybrid retrieval: finding the right passage1 part · 21 min
- Module 9: Hybrid retrieval: finding the right passage21 min
12Module 10: The agent, a graph with tools and approval1 part · 22 min
- Module 10: The agent, a graph with tools and approval22 min
13Module 11: Model routing: two models, one answer1 part · 22 min
- Module 11: Model routing: two models, one answer22 min
14Module 12: Guardrails and personal data masking1 part · 23 min
- Module 12: Guardrails and personal data masking23 min
15Module 13: Prompt versioning: changing the assistant safely1 part · 23 min
- Module 13: Prompt versioning: changing the assistant safely23 min
16Module 14: Evaluation: measuring quality before customers do1 part · 25 min
- Module 14: Evaluation: measuring quality before customers do25 min
17Module 15: Observability: traces, metrics, logs and cost1 part · 23 min
- Module 15: Observability: traces, metrics, logs and cost23 min
18Module 16: The frontend and running it locally1 part · 24 min
- Module 16: The frontend and running it locally24 min
19Module 17: Containers, the Dockerfiles and the local stack1 part · 23 min
- Module 17: Containers, the Dockerfiles and the local stack23 min
20Module 18: Deploying to Google Cloud, with CI/CD1 part · 23 min
- Module 18: Deploying to Google Cloud, with CI/CD23 min
Eighteen illustrated modules that take a working system apart. The system is in ../source-code/: a digital bank whose AI assistant answers customers from the bank's own documents and the customer's own data, and moves money only after the customer approves. Admins upload the documents and watch quality.
Who it is for: anyone who knows a little Python and wants to understand, rebuild or extend a production-style AI application, from beginners to working engineers. Each module says what to read first.
How each module is laid out
Every module explains one part of the system in the same order, so you always know where to look.
Source code
The complete, runnable project as one zip: the code the modules explain, with its tests and the files to run and deploy it.
The source code download comes with buying this project.
Submit your build
When the last milestone runs, hand in the repository and three lines about it. That is what the project page will show for you, and what you can send with a job application.
Sign in to submit your build: the repository link and three lines about it.
Reviews
to review this project once you have finished it.