Projects

TWITTER TOPIC MODELLING

I scraped text data off Twitter using Tweepy to discover the hidden topics and relationships between each tweet with the use of Topic Modelling; an unsupervised algorithm. Once a topic a discovered, a suggested link is given to read more about it. Project link

Employee Analysis with PostgreSQL

This project demonstrates the use of SQL for end-to-end employee data analysis; including preprocessing; feature engineering; and exploratory analysis. It applies advanced SQL techniques such as joins; subqueries; views; and foreign keys to uncover insights into employee demographics; salaries; positions; marital status; and branch performance across multiple countries. The analysis highlights key workforce trends; such as salary disparities by gender; position-based career longevity; and country-level employment patterns; while showcasing strong data wrangling; transformation; and insight generation skills. Project link

Research Assistant Using Cohere Reranker with Active Loop's VectorStore for QA with Sources

This project implements a research assistant using RAG; LangChain; Hugging Face models; Cohere’s reranker; and Active Loop’s Deep Lake as a vector store. It enables users to query stored research articles and receive concise; source-backed answers tailored to their questions. The system is designed to help researchers quickly extract relevant information from academic papers; while showcasing skills in vector databases; retrieval-augmented generation; and QA pipeline development.

Project link.

Publication Link