Practical Data Science on the AWS Cloud (Specialization)

Project GitHub Link

Keywords - AWS Cloud, AutoML, MLOps, A/B testing, End-to-End Model Deployment, NLP, BERT, Statistical Biases

Pre-requisites - Deep Learning & AWS Cloud Technical Essentials specialization, Machine Learning, Python, Statistics

Certificates - certificate1 , certificate2, certificate3, certificate4

This specialization focused on -

1. Preparing data, detecting statistical biases, performing feature engineering at scale to train models, evaluate and fine tune models with AutoML.

2. Building, deploying, monitoring and operationalize end-to-end machine learning pipelines.

3. Storing and managing ML features using feature store. Debugging, profiling, tuning, & evaluating models while tracking data lineage and model artifacts.

4. Building data labeling and human-in-the-loop pipelines to improve model performance with human intelligence.
   

The repository contains following notebooks completed as part of the specialization -

PART-1

1. Registering and visualizing dataset

2. Detecting Data Bias with Amazon Sagemaker Clarify

3. Training ML models using Amazon Sagemaker Autopilot
   

PART-2

1. Training a text classifier using Amazon SageMaker BlazingText built-in algorithm

2. Feature transformation with Amazon SageMaker processing job and Feature Store

3. Building a SageMaker Pipeline to train and deploy BERT-Based text classifier
   

PART-3

1. Model Optimization using automatic model tuning

2. A/B testing, traffic shifting and autoscaling

3. Data labeling and human-in-loop-pipelines with Amazon Augmented AI