Machine Learning & Data Science Projects

Explore my portfolio of machine learning and data science projects spanning demand forecasting, recommender systems, graph neural networks, and optimization algorithms.

E-commerce Recommender System (MLOps)

A production-ready end-to-end recommender system for e-commerce platforms featuring:

  • Data & model versioning with DVC
  • Experiment tracking & model registry using MLflow
  • Data & concept drift detection with Evidently
  • Deployable API/UI with FastAPI/Streamlit
  • CI/CD automation with GitHub Actions

Technologies: Python, DVC, MLflow, Evidently, Implicit, FastAPI, Streamlit, GitHub Actions


Supply Chain ML Pipeline

An end-to-end machine learning pipeline for supply chain optimization featuring:

  • Automated ML pipeline orchestration with Apache Airflow
  • Demand forecasting models for inventory planning
  • Inventory optimization algorithms
  • Workflow scheduling & monitoring
  • Scalable data processing

Technologies: Python, Apache Airflow, Machine Learning, Time Series Forecasting, Supply Chain Analytics


Graph Neural Networks

Research and implementation of Graph Neural Network (GNN) architectures for:

  • Node classification and link prediction
  • Graph-level learning tasks
  • Graph representation learning
  • Research & experimentation framework

Technologies: Python, PyTorch Geometric, Deep Learning, Graph Theory


Ad Campaign Optimizer

A machine learning system for optimizing digital advertising campaigns featuring:

  • Budget allocation optimization algorithms
  • Audience targeting models
  • Campaign performance prediction
  • ROI maximization strategies
  • Real-time campaign analytics

Technologies: Python, Machine Learning, Optimization Algorithms, Marketing Analytics


Time Series Forecasting

Advanced demand forecasting models developed for supply chain optimization, including:

  • ARIMA models for stationary time series
  • LSTM and Bayesian LSTM networks for non-stationary data
  • Generative Adversarial Networks (GANs) for demand simulation
  • Attention mechanisms for complex patterns

Technologies: Python, TensorFlow, PyTorch, Keras, scikit-learn, pandas, numpy


Amazon Product Reviews Analytics

A full-stack machine learning web application that classifies Amazon product reviews using natural language processing:

  • Topic modeling using Latent Dirichlet Allocation (LDA)
  • Web scraping and data pipeline for review collection
  • Product recommendation based on review keywords
  • Deployed web application using Flask and Heroku

Technologies: Python, Flask, Heroku, LDA, NLTK, Beautiful Soup