My Portfolio

Tarot Reader

I've been interested in tarot card readings for quite a while. This project was my attempt to create an automated tarot card reading app using ChatGPT and a Flask web server.

A user can input a query to the form, and select their reading type (typically three-card reading) after which they send their query to the server. Using a secret technique that draws cards from a tarot deck on the server that has the most connection to the user, ChatGPT is offered their query and the chosen cards.

ChatGPT will generate a response based on this data and return it to the user. As the cards are typically open to interpretation, some interesting and insightful responses can show up that, I have been told, are often deeply relevant to the querent.

It runs on the same Flask app as this website on AWS, which was a challenge to set up as it has a different domain name "akashika.org". All HTTP requests hitting the AWS server bound for akashika.org are routed via a forward proxy on nginx to the endpoint /tarot-reader on the Flask web server. This mirrors the effect of hosting on a separate server without the overhead of two Flask apps.

I had fun coding it and would encourage anyone to give it a try and see what you think.

Click Here

Portfolio Manager Lite

This is an example portfolio management application that was made to manage stock market portfolios with real-time data. It runs as a client version on Flask which can be embedded into other Flask apps (as I have done on this server.)

The core of the app is the creation of "portfolios". These portfolios can hold stocks, purchased and sold via the trade URL, and report their financial performance over a given period.

The stock market data I collect from polygon.io is stored in a MySQL server on Amazon RDS. The app is configured to pull data from this website, so regardless of where the client version of the app is running, it will get the same price data.

If you want to try installing the client (or to view some of the back-end code in Python) check out my GitHub repository at https://github.com/laked0601/PML-Pubdivc.

Click Here

Tax Calculator

This is an example of a tax calculator I coded in HTML/CSS/Javascript. Although not my strong suit for languages, it was interesting to put together the logic for determining tax levels and researching tax law across multiple jurisdictions.

My background is in accounting, so the overall rules for determining income tax levels were quite familiar. However, for St Pierre and Miquelon the tax law is not very well documented (for example) so I had to find a pdf from the Government website and translate directly from French.

Click Here

Guernsey Registry Data

This was an investigative project to see how the API for the Guernsey Registry functioned and if I could get some interesting data from it.

I scraped the majority of the registry, stored the data in MySQL, and performed a series of analyses using Power BI. I had wanted to create an interactive Power BI dashboard in HTML but, frustratingly, a license is required from Microsoft to do this. A pdf had to suffice as such.

I believe this is an excellent example of where having developers is critical, as the insights found from getting the entire data set would not have been accessible to an individual without the technical know-how of how a rest-based API functions.

Click Here

365GPT

This is a macro file for Excel that embeds ChatGPT as a function in Excel. It's particularly useful for gauging things like customer sentiment or doing data analysis where traditional programming methods wouldn't work.

The repository link above includes a tutorial to get it set up and where to get your API key from.

Click Here

CompaniesHouseGPT-Public

This was an experiment I tried to test the effectiveness of ChatGPT in analyzing financial statements from the UK Companies House Registry.

It involved scraping a few thousand PDFs, OCRing and storing the text, and then exposing ChatGPT to this corpus of financial data.

I posed the question "Who was the auditor (if any) of the financial statements" and the results were interesting. 320 were audited, and 680 were not. The top named auditors by companies were KPMG (19), Deloitte (15), PWC (12), JSA Services Limited(9) and MMO Limited(9).

Overall though it isn't shocking that the top audit firms would be the largest auditors I think it shows the usefullness of the technology.

Click Here