Fortuna

Building agent
Received predictions
Using the Cloud build to invoke the model daily
Forecasting using Datasets
Configuring the Endpoints for predications
Specifying the type of file to upload
Fortuna User Dashboard
Agents Knowledgebase
Using the batch prediction as function
Fortuna Site Login Page
Fortuna Site Profile Page

Inspiration

Just lately I stumbled upon a video on YouTube explaining 'How to build forecasting models with Vertex AI'. In it Mrs. Priyanka Vergadia from Google Cloud Tech explained how we can create MLOps pipelines with ease using VertexAI and the Google developers Suit. I have experience with stock market predication models so I immediately thought of using the information from the video and create my own market forecasting model.

What it does

The model forecasts the predicted stock values for any and all companies whose data can be accessed through the yfinance module which uses the information from yahoo finance's ticker data. I use this forecasted data and use it as knowledge base for Agents trained in VertexAI. These agents are deployed on our WebApp linked below

How we built it

I built it completely on Low/No code platforms . I used VertexAI to train the model based on the performance of the S&P500 data, since it represents a stock thats always been growing and a good indicator of overall market preformance. I fed this data to managed datasets using the forecast function for tabular data and AutoML to automatically train the data by simply specifying the Target Columns and Timestamps. Once I have trained the model we can deploy it on an endpoint which we can call through API's, every predicted batch is saved to the Batch Predications. I then create an Agent, and give it access to some of the biggest financial adviser sites as well as our latest batch predications to give the client informed and calculated insights to the market. This as agent is deployed on the Fortuna site and is used for various uses such as recommending purchases, selling stock that is going to depreciate or general insights for a persons portfolio.

Challenges we ran into

Having no prior experience in MLOps or Web Design or Application, I found myself in a time crunch , but due to the availability of plenty resources and tutorials by Google and it's affiliated branches, I ended up picking up new tools, skill very fast.

Accomplishments that we're proud of

I are particularly proud of the time in which this project conceived and how fast we managed to build it. I found out about the hackathon only on the 1st of May but with the help of tools provided by Google's Vertex AI we managed to complete the project within stipulated time. As well as the fact that everything in the project is AI generated from the little code required to the logos and the narration voice you hear on the project video.