MLOps Projects

Insurance Costs Prediction

Created a predictive model for Insurance Costs Prediction, implementing EDA and Tensorflow regression model with hyperparameter tuning to optimize metrics. Utilized Comet for model tracking, several model registry and deployment pipelines and data/model quality monitoring. Implemented CI/CD pipelines for seamless integration and deployment on AWS

Stroke Prediction

Implemented a classification model for Stroke Prediction, integrating EDA and implementing an XGBoost model with threshold-moving for improved prediction of the minority class. Tracked the experiments in Comet ML, and served the model on Flask, using Docker for containerization. The model was deployed and registered on AWS, using conditional regitry (RO-AUC Threshold)

Taxi Rides Prediction

Implemented a predictive model for Taxi Rides Prediction, integrating EDA and implementing a neural network model with multiple dense layers and batch normalization for enhanced performance. Leveraged Prefect for workflow management, FastAPI for API integration, Docker for containerization, and deployed the solution on Google Cloud Platform (GCP)

Music Clustering

Developed a end-to-end project for clustering a Spotify music dataset, integrating EDA, KMeans, and PCA. Utilized FastAPI for API development, Docker for containerization, implemented CI/CD pipelines, and deployed the solution on AWS, creating additionally an interactive Streamlit app

Food Prediction

Implemented a predictive model for Food Prediction, integrating EDA and a neural network transfer learning model with multiple dense layers and augmenation for enhanced performance. Leveraged FastAPI for API integration, Docker for containerization, and deployed the solution on Google Cloud Platform, creating additionally an interactive Streamlit app

Birds Classification

Developed a Computer Vision project, leveraging PyTorch EfficientNet models for enhanced accuracy. Extended the project to include deployment in Gradio

Car Price Prediction

Created a predictive model for Car Price Prediction, implementing EDA and diverse regression models with hyperparameter tuning to optimize performance and accuracy. Utilized MLFlow for model tracking, Prefect for workflow management, Flask for API development, Docker for containerization, Grafana for monitoring, and Terraform for infrastructure provisioning. Implemented CI/CD pipelines for seamless integration and deployment on AWS