Financier

AI-powered personal finance management API

⚠️ Work in progress

As much as I love to work on this project fulltime, I still have a job that I love.

🧠 What is it?

Basically a glorified bookkeeper, financial coach, and financial manager. I wanted to leverage the use of OpenAI models to create an assistant that will:

• keep track of my expenses
• generate reports
• remind me of a money-related event
• make me feel good about my financial health (or not)

Features

1. 📃 Create, list, read, update, delete income and expense items.
2. 📸 Upload a photo of a receipt and let the assistant book it for you.
3. 📈 Generate daily, monthly, yearly reports of your finances.
4. 💪🏽Set budgets (daily and weekly).
5. 📨 Receive emails about alerts and notifications.
6. ⛰️ Set financial goals.
7. 👨🏽 Ask the assistant about your finances.

Installation

1. Use poetry for the project.
2. Define .env with the following contents:

STAGE=<stage>
APP_NAME=<name>
PYTHON_PATH=${workspaceDirectory}

Creating the Lambda layers

All the requirements are found in the financier/functions_stack/layers folder. We are using pydantic and there is a special build needed for it to make it working with Lambda.

pip install --platform manylinux2014_x86_64 --implementation cp --python-version 3.12 --only-binary=:all: --upgrade pydantic --target <directory>

Make sure to update this when you are using a new python package.

Running Local Development

1. Enable your environment.
2. Go to financier/functions_stack/
3. Run local FastAPI development server,

fastapi dev fastapi_lambda.py

Quality

The project uses: isort, ruff, and black for code formatting. I've created a .bat file to run the necessary steps.

runner.bat

This will be improved in the future as I am trying to make tox work for the project.

🛣️ Roadmap

Phase 1

Initial setup.

• Initialize project structure
• Setup project
• Setup tests

Phase 2

Simple proof of concept for data management.

• Data modelling
  • Expense
• Database stack
  • DynamoDB setup
• API stack
  • Create
  • Read
  • List
  • Update
  • Delete
• Lambda stack
  • Setup lambda layers
  • Create lambda layer creation script
  • Create
  • List
  • Read
  • Update
  • Delete
• Convert function creations to use YAML loading

Phase 3

Securing our application.

• Stardardize IAM roles
• Cognito setup
  • Identity pools
• API Gateway authorizer

Phase 4

Setup the assistant

• Write OpenAI model code for the assistant.
• Interface for the Financier API and the assistant API.

Phase 5

Assistant features

• Tools for the assistant
  • Reports generator
  • Chat
  • Photo manager

Phase 6

More assistant features

• Budget
• Alarms
• Goals