my work

Finetuning AudioLDM for Higher Quality Music Generation

Fine-tuned AudioLDM, a text-to-audio model, for high-quality music generation using the MusicQA dataset. Conducted programming and experimental validation to optimize the model's performance.

#VAE #LDM #GEN AI #Conditional Generation

META Ads Reporting and Analysis Tool

Meta Ads Reporting and Analysis Tool—a cloud-powered pipeline that transforms raw META (Facebook/Instagram) ad exports into structured insights and business-ready reports.

#Python #postgresql #AWS

Fine Grained Classification of Knives using CNNs

Using Convolutional Neural Networks to perform fine grained classification of types of knives while exploiting transfer learning using different EffecientNet and ResNet models and contrasting their performance. Ensemble models are also explored.

#Python #Pytorch #NumPy #Matplotlib #CNN

Vehicle Re-Identification

Developed architectures such ass CNNs and Transformers to match vehicle images from non-overlapping camera views. Used the VeRi dataset.

#CNN #TRANSFORMERs #comp vision

Training Diffusion Models for Image Generation

Trained a diffusion model for image generation using the CIFAR-10 dataset. Tuned hyperparameters and evaluated model performance using FID and Inception Score metrics.

#DDPM #GEN AI #comp VISION #UNET2D

Sequence Classification for Abbreviation and Long Form Detection

Built multiple sequence classifiers for detecting abbreviations and long forms using varous techniques to inestigate effectiveness of each.

#NLP #PLOD DATASET

Rossmann Sales Prediction

This repository hosts an end-to-end data science project for predicting daily sales of Rossmann stores using historical data, promotions, and holidays. It features a complete pipeline from data ingestion (via AWS S3) to preprocessing, time-series modeling (XGBoost, Scikit-learn), and interactive visualization (Power BI/AWS QuickSight).

#pandas #python #xgboost

Q and A Chatbot

A simple Question and Answer chatbot written in Python used to answer questions from a larger database

#NLP #NLTK

Bird Classifier

A web app with bird classification capabilities built using ML5 for machine learning and p5 in Javascript

#ML5 #Machine Learning #JavaScript #p5

Incorporating News Sentiment in Portfolio Optimization with Return Prediction

Using Machine Learning methods to predict stock prices and output optimal portfolios using technical indicators and news sentiment as features.

#Python #R #NumPy #Pandas #sklearn

Web Development

Keystone website

#WordPress #Elementor Pro

1st Alliance Website

#WordPress #Elementor Pro

Roots Website

#WordPress #Elementor Pro

1st Class Website

#WordPress #Elementor Pro

We Are Urban Website

#Shopify #Ecommerce

Get in touch!