Implementing & Training a Transformer

Implemented and trained an encoder-decorder seq-2-seq transformer for language translation

Store Chain Sales forecasting with Machine Learning

Created and evaluated multiple forecasting models (Prophet, XGBoost, CatBoost, LightGBM) to predict daily sales for a restaurant chain across 163 locations, using daily iterative forecasting

2024 U.S.A Federal election outcome prediction & Visualization

Wrangled a 53 column dataset from several sources in Python (pandas) with feature engineering. Used k-nearest neighbours, Logistic Regression, & Random Forest models to predict outcomes on the county-level in the 2024 US Federal election. Built an interactive data visualization in Streamlit as well.

Fine Tuning an open source LLM for dialogue summarization

Fine-tuned Google's flan-t5 on a large-scale dialogue summarization dataset using techniques like LoRA and parameter-efficient fine-tuning, reducing inference errors by 20%.

Retrieval Augmented Generation Workflow

Built a Retrieval-Augmented Generation (RAG) pipeline by integrating LangChain, ChromaDB, and gpt-5, enabling context-aware, domain-specific text generation.

Loan Approval Prediction

Used Logistic Regression to determine whether or not a client will default on their loan. Used binning to transform continuous variables into categorical variables for ease of understanding and to be put onto a simple scorecard.

COVID-19 Cases & Vaccinations Data Analysis

Extracted statistics over time about the COVID-19 pandemic with aggregate functions in SQL. Built dashboards that generate insight into the dataset with custom graphs and filters for drill-down.