Live at

💸 banqstream.com

Background

This was a project I created to exercise some of the new trends on LLMs. I was particularly interested in LLM’s capacity to extract information from raw inputs and further transform it either by grouping similar things ( embeddings > ANN > etc) to transforming and proposing new meaning to data (assigning automatic labels for example).

Subjects to explore

  • Basic LLM functionalities
  • Embedding + ANN + clustering
  • LLM finetuning

Quick overview

This is an attempt to create a finance/budgeting assistant. This will be using LLMs to parse raw text coming from account statements and feeding this into a database used for further analysis. Among the features on this app:

  • Parse raw account transactions
  • Assign categories based on transactions names (try to make this as consistent and flexible as possible)
  • Split big inputs from customers that might exceed model max tokens limitations (i.e 4096 etc)
  • Simple dashboard for data visualization
  • Enable user to talk to its own data through chat UI

To-Do

  • Design a dedicated UI, possibly with fastAPI and React.
  • Make data parsing faster (maybe experiment with different models?)
  • Experiment with fine-tuned models for the AI assistant.
  • Enhance the UI with added functionalities for editing transactions.