Build Financial Software with Generative AI (From Scratch) you own this product

Christopher Kardell and Mark Brouwer

April 2025
ISBN 9781633436626
432 pages

Included with a Manning Online subscription

printed in black & white

available in Russian, Simplified Chinese

catalog / Data Science / Deep Learning / Generative AI

table of content

Part 1 FinTech and the Automated Clearing House

1 Exploring FinTech and generative AI

1.1 The Futuristic FinTech project

1.2 ACH dashboard

1.2.1 How important is the ACH?

1.3 SAFe Agile

1.4 The ACH dashboard

1.4.1 Tools of the trade

1.5 A first look at generative AI

1.5.1 A better rubber duck

1.5.2 Hello, Check 21!

1.5.3 Help! This code is not working!

1.5.4 Help! I forgot the syntax!

1.5.5 Did I just give away my data?

1.5.6 Privacy and security

1.5.7 Staying up to date

Summary

2 Parsing payments

2.1 Modernizing our legacy software

2.2 Understanding the ACH

2.3 Parsing an ACH file

2.3.1 Asking ChatGPT to parse an ACH file

2.3.2 Parsing an ACH file with Copilot

2.3.3 Generative AI: Trust but verify

2.4 Automated testing

2.4.1 Testing the number of records read

2.4.2 Parsing the records

2.4.3 File header record (type 1)

2.4.4 Batch header record (type 5)

2.5 Entry detail record (type 6)

2.5.1 Addenda record (type 7)

2.5.2 Batch control record (type 8)

2.5.3 File trailer record (type 9)

2.5.4 Passed!

2.6 The not-so-happy path

2.6.1 File rejection

2.6.2 Batch rejection

2.6.3 Entry rejection

2.7 Interpreting the code

2.7.1 Where’s my switch statement?

2.7.2 Type hints

2.7.3 Secure coding

Summary

Part 2 A path to the minimum viable product

3 Getting started with Docker

3.1 Where to begin

3.2 Creating a docker-compose.yml file

3.2.1 Creating a CloudBeaver container

3.2.2 Create a PostgreSQL container

3.2.3 API container

3.2.4 Web server container

3.3 Connecting our containers

3.3.1 Connecting to the database

3.3.2 Troubleshooting PostgreSQL and FastAPI

3.4 Calling our API

3.4.1 Troubleshooting our web site

3.5 Container security

3.6 Optimizing Docker

3.6.1 General maintenance

3.6.2 Optimizing image size

3.6.3 Optimizing build time

3.7 Removing the hardcoded username/password

3.8 Health checks

Summary

4 APIs: The piping between our components

4.1 API-driven design

4.2 Our APIs

4.3 Building our APIs with FastAPI

4.3.1 Retrieving files with GET /files

4.3.2 Adding a file with POST /files

4.3.3 Finishing the rest of the APIs

4.4 Adding our API to Docker

4.4.1 Preparing the project

4.4.2 Docker setup

4.5 Documenting our API

4.5.1 Swagger

4.5.2 Redoc

4.5.3 What is so great about API documentation?

4.6 Managing our APIs

4.6.1 Benefits of API management

4.6.2 Adding WSO2 API manager to our project

4.6.3 OpenAPI Version

4.6.4 Create an API in WSO2

4.6.5 Troubleshooting

4.7 Testing with Postman

4.8 Load testing with Locust

4.9 Improving our code

4.9.1 Using a prefix and versioning

4.9.2 Adding Pydantic

4.9.3 Validation with Pydantic

4.9.4 Troubleshooting Pydantic

Summary

5 Storing our ACH files

5.1 Designing our database

5.2 Using SQL directly

5.2.1 Adding records to the ach_files table

5.3 Storing the file header record

5.3.1 Using generative AI

5.3.2 Full example

5.4 Storing the rest of our ACH records

5.4.1 Storing ACH files challenge: Lessons learned

5.5 Storing exceptions

5.6 Uploading an ACH file

5.7 Storing records with integrity

5.8 Using Pydantic

5.9 Lessons learned

5.10 Coding changes

5.10.1 Creating Pydantic schema for the unparsed ACH records

5.10.2 Creating Pydantic schema for the parsed ACH records

5.10.3 Unit test changes

5.11 Design and different approaches

Summary

6 Taking the next step with Next.js

6.1 Getting started with Next.js

6.2 Creating an upload page

6.2.1 Testing our upload page

6.3 Refactoring our navigation

6.4 Converting recent order to recent ACH uploads

6.5 Mocking our ACH uploads data

6.5.1 Adding WireMock to Docker

6.5.2 Setting up mappings in WireMock

6.5.3 Accessing endpoints with axios

6.6 Expanding the UI with an ACH batch view

6.6.1 Defining our WireMock data for the ACH batch view

6.6.2 Updating the dashboard to take us to the batches screen

6.6.3 Displaying batch details

6.7 Creating a ScatterPlot component

6.7.1 Scatterplot component

6.7.3 Scatterplot custom ticks

6.7.4 Using colors for the chart

6.8 Deploying the dashboard in Docker

Summary

7 Our minimum viable product

7.1 Which minimum are we talking about?

7.2 Preparing for the MVP

7.3 Building out the /files APIs

7.3.1 Testing the API

7.4 ACH batches APIs

7.4.1 Adding the /files/id/batches API

7.4.2 Supporting ACH batches in the dashboard

7.4.3 Uploading files

7.5 Putting the puzzle together

7.5.1 Cleaning up the configuration files

7.5.2 Ensuring APIs are accessible

7.5.3 Programming challenges

7.5.4 Exploring the dashboard

7.6 Gotta have more files

7.7 Adding a basic login page

7.8 Adding TLS to our dashboard

7.9 Testing with Playwright

7.10 MVP wrap-up

7.10.1 Troubleshooting the MVP

Summary

Part 3 Beyond the minimum viable product

8 Exceptions

8.1 Planning enhancements

8.2 Exceptions and warnings

8.3 Creating ACH files with exceptions

8.3.1 Creating ACH files with file-level exceptions

8.3.2 Creating ACH files with batch-level exceptions

8.3.3 Creating ACH files with entry-level exceptions

8.3.4 Clean-up of our file creation syntax

8.4 Expanding the ACH processing for exceptions

8.4.1 ACH exceptions in the database

8.4.2 Recovery options

8.5 Updating the ACH parser

8.5.1 Creating unit tests for the exceptions

8.5.2 Supporting the exceptions in our parser

8.5.3 Validation in Pydantic

8.6 Updating the APIs

8.6.1 Retrieving an unparsed record

8.7 Insights into exceptions with the UI

8.7.1 Showing all the exceptions

8.7.2 A visual cue for exceptions

8.8 Additional UI changes

8.8.1 Creating an exceptions component

8.8.2 Batch statistics component

8.8.3 Adding the application

8.8.4 Making our UI more responsive

Summary

9 Searching and auditing

9.1 Sprint planning

9.2 Searching ACH transactions

9.2.1 BDD testing for searching

9.2.2 Search API

9.2.3 UI search page

9.2.4 Integration testing with Playwright

9.3 Auditing user interactions

9.3.1 Database design

9.3.2 TDD testing for auditing

9.3.3 Auditing logic

9.4 Using middleware for our logging

9.5 Viewing the audit log

9.5.1 Creating the page

9.5.2 Creating the page components

9.5.3 Adding the APIs for the audit log

Summary

10 Company information

10.1 Sprint planning

10.2 Company Information

10.2.1 Database design: Company information

10.2.2 Using generative AI to generate mock data

10.2.3 API design

10.2.4 User interface

10.3 Company dashboard

10.3.1 Navigation for the company dashboard

10.3.2 Creating ACH files for our companies

10.4 Company Information component

10.5 Recent Batches component

10.5.1 Implementing the recent batches API

10.5.2 Displaying recent batches

10.6 Expected Files component

10.6.1 Storing expected files

10.6.2 Implementing the expected files API

10.6.3 Displaying the expected files

10.7 Company Limits component

10.7.1 Storing the company limits

10.7.2 Implementing the company limits API

10.7.3 Displaying the company limits

Summary

11 International ACH transactions and OFAC scanning

11.1 Sprint planning

11.2 International ACH transactions

11.2.1 IAT batches: An overview

11.3 Creating IAT batches

11.3.1 Database changes

11.4 IAT record parsing

11.5 IAT file processing

11.5.1 Unit testing

11.5.2 Updating file processing

11.6 Effects on the dashboard

11.6.1 The get_batches method

11.6.2 Batch entries

11.7 OFAC scanning

11.7.1 Sanctioned individuals and countries

11.7.2 Scanning for individuals

11.8 Application programming interface

11.8.1 Unit test

11.8.2 Creating the endpoint

11.8.3 Finishing the API

11.9 User interface

11.9.1 The OFAC page

11.9.2 OFAC components

11.9.3 OFAC results

Summary

12 Where to go from here

12.1 Backend processing

12.1.1 Asynchronous processing

12.1.2 ORM frameworks

12.1.3 Application programming interfaces

12.2 Infrastructure

12.2.1 Docker

12.2.2 Multitenant environment

12.2.3 Continuous integration and delivery

12.2.4 Testing

12.2.5 Database

12.3 User interface

12.3.1 Charts and forms

12.3.2 Configurable dashboards

12.3.3 Mobile version

12.4 Finance

12.4.1 Accounts and customers

12.4.2 ACH origination and returns

12.4.3 Positive Pay

12.4.4 Service charges

12.5 Final thoughts

Summary

Overview

1 Exploring FinTech and Generative AI

This chapter introduces FinTech as the application of technology to improve how money is handled, tracing its roots from early tools to today’s digital payments, crowdfunding, gig platforms, and cryptocurrencies. It sets expectations for a full-stack learning journey: readers will work across UI, APIs, databases, and backend services while modernizing a real-world style payment tool for a fictional company, Futuristic Fintech. The project centers on the Automated Clearing House (ACH), a cornerstone of US payments, using it as a concrete context to explore systems design, data flow, and the practical realities of building financial software.

The chapter frames the modernization of an ACH Dashboard as a typical enterprise initiative, explaining why products are refreshed (compliance, cost, design alignment, repositioning) and how such efforts unfold. It outlines ACH’s prominence and interaction with international counterparts, the rise of Same Day ACH, and the broader push toward real-time payments. Work is organized with SAFe Agile concepts like PI Planning, iterations, and epics/stories. Clear problems in the legacy interface (file uploads, reporting, access constraints, and inconsistent styling) motivate a target architecture featuring PostgreSQL for analytics-friendly storage, a Next.js UI with visualizations, Python replacing COBOL services, and Dockerized components, supported by common developer tooling with flexibility to substitute equivalents.

The chapter then offers a pragmatic first look at Generative AI as a “better rubber duck” that can brainstorm, scaffold code, add type hints, debug logic, and accelerate tasks from Python snippets to simple D3 visualizations. It stresses prompt engineering and problem formulation, but equally emphasizes verification and accountability for generated code. Crucial cautions include safeguarding sensitive data (NPI/PII) and proprietary code, respecting ownership and policy constraints, and recognizing model currency and version differences. The takeaway is a balanced approach: Generative AI can materially boost productivity in financial software development when paired with disciplined validation, security awareness, and sound engineering practices.

Building the ACH Dashboard

Loading a file through the ACH Dashboard

Year-over-year comparison of various payment methods

PI Planning Template in Miro

Legacy ACH Interface

Modern ACH Dashboard

ACH Dashboard Initiative

Simple Bar Chart

Bar Chart with axis and labels

Highlighting and Hovering in our Bar Chart

Summary

FinTech encompasses a wide range of financial technologies, historically from cash registers to modern-day applications, including crowdfunding, gig service apps, P2P payments, and cryptocurrencies.
The book targets full stack developers in FinTech, covering end-to-end application development including UI, backend processing, and troubleshooting through real-world examples.
We introduced the SAFe Agile framework, highlighting its widespread use in FinTech companies and the significance of Program Increment Planning in managing software development projects.
We emphasized modernizing legacy systems using new tools and frameworks such as PostgreSQL, Next.js, Docker, and Python, and considers the impact of such changes on users and businesses.
Generative AI is seen as an evolution of the "Rubber Duck" debugging method, providing active feedback and ideas.
Generative AI uses Large Language Models and is integrated into various industries for increased efficiency.
Generative AI assists in coding, debugging, syntax recall, and generating code snippets across programming languages.
Readers should be cautious about relying on AI for up-to-date information due to potentially outdated training data.

FAQ

What is FinTech?

FinTech is the use of technology to improve how money is handled. It powers faster, more convenient, and often new ways to pay, save, invest, and move funds—from crowdfunding and gig-economy payouts to P2P apps like Venmo and traditional banking, investing, and credit products.

What is the Automated Clearing House (ACH) and why is it important?

ACH is a U.S. network used to move money between bank accounts (for example, payroll direct deposits, bill payments, and account-to-account transfers). It has grown rapidly as checks decline, and enhancements like Same Day ACH have reduced settlement times from days to hours.

What will I build in the ACH Dashboard project?

You’ll build a full-stack ACH Dashboard that loads ACH files, parses and stores them in a relational database, exposes data via APIs and backend processing, and presents insights in a modern UI. The project emphasizes analytics, potential fraud monitoring, and a consistent look-and-feel aligned with enterprise standards.

What problems exist in the legacy ACH interface and how will modernization address them?

Key legacy limitations: no direct file uploads, minimal reporting/analytics, limited access due to server constraints, and outdated styling. The modernization initiative targets:

PostgreSQL for structured storage and BI analytics
Next.js for a production-ready React UI with visualizations
Docker to containerize components
Python for new backend services

Which tools and technologies are used in this book?

The stack centers on PostgreSQL, Next.js, Docker, and Python. Common developer tools include JetBrains PyCharm/WebStorm or VS Code, JetBrains Qodana for code review, and AI assistants like ChatGPT and GitHub Copilot. You’re encouraged to substitute equivalents (e.g., MySQL, Java) if preferred.

How does SAFe Agile and PI Planning fit into the project?

The team plans work using SAFe. A Program Increment (PI) spans about 8–12 weeks and includes several sprints. During PI Planning, work is shaped into epics and user stories, dependencies are identified, risks are ROAMed (Resolved, Owned, Accepted, Mitigated), and a confidence vote confirms the plan.

How will Generative AI be used in the project?

Think of it as a “better rubber duck.” Generative AI helps summarize regulations, bootstrap code, troubleshoot bugs, and accelerate learning across the stack (e.g., Python, SQL, HTML/D3 for charts). It can speed up iteration if you provide clear prompts and validate outputs.

What are the main risks and best practices when using Generative AI in FinTech?

Key considerations:

Privacy and security: avoid sharing NPI/PII and proprietary code; review tool policies
Ownership/IP: understand ToS/EULA before integrating generated code
Verification: test, review, and maintain code you adopt
Prompt quality: be clear, concise, and provide examples
Security guidance: see OWASP’s GenAI resources at https://genai.owasp.org/

How does an ACH file flow through the modern dashboard?

At a high level: a user uploads an ACH file; the system parses it and writes records to a relational database; backend services and APIs expose the data; and the UI visualizes and filters transactions for operations, analytics, and decision-making.

What skills are expected from readers and what will they gain?

You should know basic programming constructs (loops, conditionals) but need not be tied to a specific language. You’ll learn full-stack problem solving—tracing issues from UI events through APIs to server and SQL—plus experience with modernization, Agile workflows, and practical, safe use of Generative AI.

pro $24.99 per month

access to all Manning books, MEAPs, liveVideos, liveProjects, and audiobooks!
choose one free eBook per month to keep
exclusive 50% discount on all purchases
renews monthly, pause or cancel renewal anytime

lite $19.99 per month

access to all Manning books, including MEAPs!

team

5, 10 or 20 seats+ for your team - learn more

eBook

pdf, ePub, online

$47.99 $23.99

you save $24.00 (50%)

include audio $24.99 $12.49

pro $24.99 per month

access to all Manning books, MEAPs, liveVideos, liveProjects, and audiobooks!
choose one free eBook per month to keep
exclusive 50% discount on all purchases
renews monthly, pause or cancel renewal anytime

lite $19.99 per month

access to all Manning books, including MEAPs!

team

5, 10 or 20 seats+ for your team - learn more

eBook

$47.99 $23.99

you save $24.00 (50%)

include audio $24.99 $12.49

eBook

pdf, ePub, online

$47.99 $23.99

you save $24.00 (50%)

include audio $24.99 $12.49

pro $24.99 per month

access to all Manning books, MEAPs, liveVideos, liveProjects, and audiobooks!
choose one free eBook per month to keep
exclusive 50% discount on all purchases
renews monthly, pause or cancel renewal anytime

lite $19.99 per month

access to all Manning books, including MEAPs!

team

5, 10 or 20 seats+ for your team - learn more