Coding & Weekends! #
Working on small side projects have always been my hobby since college days. I try to work on different kinds of projects, like a LLM Quizzer game to a RAG based book companion app to an encryption system that is based on huffman encoding and enigma machine and finally a custom version of TFIDF model that handles class imbalance while picking features. Below are a few projects that I have worked on(both as a team and individually)
-
The scikit-learn implementation of TFIDF picks features on the basis of frequency. This approach works fine if the dataset has classes having balanced counts but can be problematic if the classes are imbalanced. Weighted Class TFIDF tackles this problem by using the same scikit learn TFIDF in a class wise manner to first select features and then use them in downstream tasks.
-
This project was my first attempt at exploring RAG systems. I specifically wanted to build a RAG system that can enable people to query books. I also showcased this project as part of a lightning talk at one of the seminars that I took in my masters.
-
As part of our minor project during my bachelors, we worked on a encryption system that combines huffman encoding with the workings of the enigma machine.
Future Works #
Recently, I have developed a fascination for creating games using Python. I aim to immerse myself in the entire end-to-end process, from designing sprites and crafting levels to enhancing gameplay mechanics for an enjoyable experience. However, before diving into large and complex projects, I plan to start by building a few simple games. One idea I am currently exploring is the development of text-based adventure games powered by Large Language Models (LLMs).