AI Certificate Program by Microsoft

In the first week, you will dive into the world of Artificial Intelligence (AI). You will get an introduction to AI, ML, DL, RL, and LLM to fully understand the scope of this program. Familiarize yourself with an extensive array of problems that can be solved using ML/DL algorithms and frameworks. Know how an end-to-end AI workflow functions.

Topics Covered:

• Practical Applications and Use-Cases of AI, ML, DL, RL, and LLMs
• Frameworks: TensorFlow, PyTorch, and Keras for implementing ML/DL models
• Concept of Evaluation Metrics: Accuracy, Precision, Recall, F1-Score, AUROC
• Interpreting Evaluation Metrics for ML models
• AI Life Cycle and stages

This week, you will learn to create and configure a data storage resource on the cloud. Grasp how to manage data in a resource, including appending, modifying, deleting, securing, uploading, downloading, renaming, and organizing it into folders. Gain the ability to query data to answer basic business queries effectively

Topics Covered:

• Overview of Database Management Systems
• Introduction to SQL and its role in data management
• Introduction to DML and DQL in SQL
• Aggregating and Organizing Data in SQL
• Overview of common in-built functions

During the week, you will learn to write complex queries to gain a deeper understanding of data. You will also learn to answer business questions. Additionally, you will run quality checks on data by documenting data profiles and conducting business sense checks. You will integrate diverse datasets to transform and prepare data for training.

Topics Covered:

• Joins - Inner Join, Left join, right Join and full Join
• Subqueries - Scalar, Row and Table Subqueries
• Set operations - Union, Intersect and Except
• Window Functions
• Custom Functions and Views

This week, you will learn the syntax and semantics of Python, including variables, data types, operators, expressions, and statements. Harness your skills in functional programming using conditional statements, loops, exceptions, and functions. Acquaint yourself with the various types of compute instances, their pricing, and their usage areas.

Topics Covered:

• Python Variables, Data Types, and Basic Operators
• Functional Programming with Conditional Statements, Loops and Lambda
• Python Data Structures: Lists, Tuples, Sets, and Dictionaries
• Overview Object-Oriented Programming in Python and use-cases
• Exploring Python Libraries and Modules for ML/AI

In this week, you will develop the ability to extract insights and identify patterns in data using statistics. You will learn to execute correlation tests to ascertain associations among data sets. Cultivate the skill of coherently presenting insights and recommendations to stakeholders

Topics Covered: 

• Implementing Descriptive Statistics for Univariate Analysis
• Perform Pearson, Spearman, and Kendall Correlation Tests
• Exploring Python's Matplotlib and Seaborn for Data Visualization
• Best Practices for Effective Visual Storytelling and Presenting Recommendations
• Dimensionality Reduction Techniques: PCA and t-SNE Using Scikit-Learn

In this week, you will acquire the ability to assemble a simple yet effective analytics and reporting workflow. Using cloud-based tools, you will learn to automate report generation, alerts, notifications, and dashboard updates. Acquaint yourself with designing a straightforward yet functional data dashboard.

Topics Covered:

• Understanding Business Analytics and Reporting Process
• Azure Alerting and Notification Services
• Creating Interactive Dashboards with Azure Power BI
• Integration of Diverse Data Sources
• Optimizing Reporting Performance and Troubleshooting

Sample Problem Statement:

E-comX, an e-commerce firm, is struggling with scattered, unstructured data hampering quick, data-driven decisions. The company requires an Azure-based solution for efficient data storage, with capabilities for complex queries, quality checks, and data integration for training purposes. Leveraging Python and compute resources, the solution should enable pattern recognition, correlation tests, and report automation. Ultimately, a simple, functional dashboard is sought to transform raw data into accessible business insights for key stakeholders for faster, more accurate decision-making.

This week, you will receive a refresher course on Machine Learning and Deep Learning to strengthen your fundamentals. This period will serve as a buffer time for those who might need additional time to finalize their projects.

This week, you will master the skill of pre-processing, training, tuning, and evaluating models for Classification and Regression. Learn the systematic procedure of hyper-parameter tuning and experimentation. Comprehend the type of resources and associated costs required for training ML models.

Topics Covered: 

• Data Preprocessing for Structured Data
• Model training and how it works for Classification and Regression Models
• Hyper-parameter Tuning 
• Model Performance Evaluation
• Costs associated with Training

This week, you will carry out pre-processing, training, and evaluating Convolutional Neural Networks (CNNs). Learn how to experiment with model parameters methodically. Also, understand the type of resources necessary and costs implicated in training Deep Learning models

Topics Covered: 

• Neural Networks Architecture
• Data Preprocessing for Image Data
• Training and Tuning Neural Networks
• Model Performance Evaluation
• Costs associated with Training Deep Learning Models

During this week, you will set up the Azure OpenAI Studio. Acquire the ability to write effective prompts for unique business use-cases. Gain an understanding of how different parameters work and influence the Language Model's responses

Topics Covered:

• Large Language Models - An Introduction
• How LLMs work - architecture and processes.
• Prompt Engineering Fundamentals 
• Parameters and Pricing
• Applying PE for point business use-cases

This week, you will master the usage of Azure OpenAI API to administer prompts on data at scale. Become proficient in evaluating prompts and responses. Learn to apply various prompt techniques such as Zero-shot, Few-shot, and CoT

Topics Covered:

• Azure OpenAI API - and how to get set up with it on Python
• Prompting Techniques: Zero-shot, Few-shot and CoT
• Evaluating Prompts
• Evaluating LLM Results
• Automating LLM Functions on Datasets

In the week, you will tackle text-based problems like Summarization and Classification. Learn to implement Retrieval Augmented Generation (RAG) for Question Answering. Deploy a user-friendly chatbot to cater to a specific use-case.

Topics Covered:

• Hugging Face - how to get setup - and how to use it?
• Summarization & Classification and their Evaluation metrics
• What are embeddings and embedding similarity measures - Cosine and Euclidean
• Vector Databases and how they are useful to implement RAG
• Deploying a chat-bot on Azure using App Services

Sample Problem Statement:

ABC Corporation is facing challenges with its customer service. They are struggling due to the inability to process queries in real-time. They need a solution that can help them in text-based problem solving, scaling up their data processing, systematically experimenting with Language Models, and having an efficient customer interaction through the deployment of a user-friendly chatbot on Azure Cloud, specifically leveraging Azure OpenAI capabilities. This will not only boost their data analytics performance but also enhance customer experience and response time.

This week, you are expected to pre-read the documentation on how to evaluate and consider costs associated with cloud solutions. Gain an overarching view of what an end-to-end Machine Learning Operations (MLOps) solution architecture entails.

In the week, you will acquire a basic understanding of version control and its implementation. Know how to set up and utilize Development and Production environments effectively. Learn to operate a simple pipeline while understanding its overall architecture.

Topics Covered: 

• Understanding and Implementing Version Control with Git
• Effective Setup and Use of Development & Production Environments
• Running a Simple Pipeline and Understanding its Architecture
• Implementing CICD on Azure
• Managing Projects and Tracking in Azure DevOps

During this week, you will acquire a basic understanding of version control and its implementation. You will learn how to set up and utilize Development and Production environments effectively. You will also learn to operate a simple pipeline while understanding its overall architecture

Topics Covered: 

• Understanding and Implementing Version Control with Git
• Effective Setup and Use of Development & Production Environments
• Running a Simple Pipeline and Understanding its Architecture
• Implementing CICD on Azure
• Managing Projects and Tracking in Azure DevOps

Sample Problem Statement:

As Y-Movies continues to expand its data-driven strategies, they face the challenge of effectively managing machine learning models from development to deployment. They are experiencing difficulties in automating tasks, maintaining versions, deploying models, and monitoring their performance. This manual and labor-intensive process has been sub-optimal, raising concerns about data quality, potential biases, and our team's capability to focus on strategic insights rather than operational tasks. Therefore, the business problem is to streamline, automate, and enhance our machine learning operations through Azure Cloud to enable efficient model management, robust model monitoring and retraining, and the creation of reliable customer-facing applications. This will eventually improve our decision-making, agility, and overall service of our streaming platform.

Building Data Products - A webinar by an Industry Expert who will give their perspective on how data products are built and integrated today into the products and services we use. You will get an idea of what lies ahead in your learning journey!