How I Passed the AI-900 Microsoft Azure AI Fundamentals Certification

A study guide for passing the AI-900 certification

Jean F Beaulieu
5 min readFeb 11, 2024

The AI-900 Microsoft Azure AI Fundamentals certification is an entry-level credential offered by Microsoft Azure that validates foundational knowledge of artificial intelligence (AI) concepts and how they are implemented using Azure services. This certification is designed for individuals who want to demonstrate their understanding of AI technologies and their practical applications in business scenarios. It covers key concepts such as machine learning, natural language processing, computer vision, and conversational AI, along with relevant Azure services like Azure Cognitive Services, Azure Machine Learning, and Azure Bot Services. Passing the AI-900 exam demonstrates proficiency in basic AI concepts and how they can be leveraged within the Azure ecosystem.

AI is a key component of today’s technological environment, altering workflows, increasing productivity, and spurring creativity in a variety of industries.

Overall, my personal motivation for pursuing the AI-900 certification revolves around my desire for continuous learning, career advancement, professional growth, and contributing meaningfully to the field of artificial intelligence.

Understanding the AI-900 Exam

The AI-900 exam is divided into five parts:

  • Describe Artificial Intelligence workloads and considerations (15–20%)
  • Describe Fundamental principles of machine learning on Azure (20–25%)
  • Describe Features of computer vision workloads on Azure (15–20%)
  • Describe Features of Natural Language Processing (NLP) workloads on Azure (15–20%)
  • Describe Features of generative AI workloads on Azure (15–20%)

Part I: Artificial Intelligence workloads and considerations

Artificial Intelligence workloads

Artificial Intelligence (AI) workloads involve tasks related to machine learning, natural language processing, computer vision, utilizing algorithms and computational resources to process and analyze vast amounts of data, enabling machines to learn, reason, and make decisions or predictions autonomously.

The different AI workloads are:

  • Computer Vision
  • Natural Language Processing (NLP)
  • Knowledge Mining
  • Document Intelligence
  • Generative AI

Guiding Principles for Responsible AI

Guiding Principles for Responsible AI include fairness, reliability and safety, privacy and security, inclusiveness, transparency and accountability, ensuring AI technologies are developed and used in a manner that is ethical, respects human rights, and promotes a positive societal impact.

The 6 Guiding Principles for Responsible AI are:

  • Fairness
  • Reliability and safety
  • Privacy and security
  • Inclusiveness
  • Transparency
  • Accountability

Part II: Fundamental principles of machine learning on Azure

Common machine learning techniques

Common machine learning techniques include supervised learning (e.g., regression, classification), unsupervised learning (e.g., clustering, dimensionality reduction), reinforcement learning (learning through interaction with an environment), and deep learning (neural networks with multiple layers), each tailored to analyze data, identify patterns, and make predictions or decisions.

The main 3 types of ML techniques are:

  • Regression
  • Classification
  • Clustering

Core machine learning concepts

In a dataset for machine learning, features are the input variables used to predict an outcome, while labels are the outcomes or predictions associated with those inputs. Training datasets train the model, and validation datasets test its accuracy on unseen data, ensuring the model generalizes well to new, similar data.

Azure Machine Learning capabilities

Automated Machine Learning (AutoML) simplifies the machine learning process by automatically selecting the best algorithms and hyperparameters, performing feature engineering, and validating models. It enables users to build high-quality models efficiently, making machine learning accessible to non-experts and accelerating the development cycle for experienced practitioners.

Part III: Features of computer vision workloads on Azure

Common types of computer vision solutions

Common types of computer vision solutions include image classification (identifying objects within an image), object detection (locating objects within an image), facial recognition (identifying individuals), and optical character recognition (OCR) (converting images of text into machine-encoded text), each enabling automated understanding and processing of visual data.

Here are the common types of computer vision solutions:

  • Image classification
  • Object detection
  • Optical character recognition (OCR)
  • Facial detection and facial analysis

Azure tools and services for computer vision tasks

Azure provides several tools for computer vision tasks, including Azure Cognitive Services Computer Vision for analyzing images and videos, Custom Vision Service for building and deploying custom image classification models, and Azure Machine Learning for more advanced model development and training. These services support tasks like object detection, facial recognition, and optical character recognition (OCR).

Part IV: Features of Natural Language Processing (NLP) workloads on Azure

Features of common NLP Workload Scenarios

In common NLP workload scenarios, key phrase extraction identifies significant terms within text, enhancing information retrieval and understanding. Named Entity Recognition (NER) classifies specific entities (e.g., names, places) for content categorization and analysis. Sentiment analysis evaluates emotional tone, aiding in customer feedback interpretation. Language modeling predicts next words, improving autocomplete features. Speech recognition converts spoken words to text, while synthesis generates human-like speech from text. Translation enables cross-language communication.

Here are the common NLP workload scenarios:

  • Key phrase extraction
  • Named entity recognition (NER)
  • Sentiment analysis
  • Language modeling
  • Speech recognition and synthesis
  • Translation

Azure tools and services for NLP workloads

The Azure AI Language service offers natural language processing capabilities, enabling text analysis, sentiment analysis, named entity recognition, language detection, and question-answering over unstructured text. The Azure AI Speech service provides speech-to-text, text-to-speech, speech translation, and speaker recognition features, facilitating voice-enabled applications. Azure AI Translator service offers real-time, multi-language text translation.

Part V: Features of Generative AI workloads on Azure

Features of generative AI solutions

Generative AI models can create novel content, including text and images, based on training data. Common scenarios for generative AI include content creation, data augmentation, personalized marketing, and simulation. Responsible AI considerations involve ensuring ethical use, preventing misuse for deceptive purposes, protecting intellectual property, and maintaining privacy and fairness by mitigating biases in generated content.

Capabilities of Azure OpenAI Service

The Azure OpenAI Service offers natural language generation capabilities, enabling the creation of coherent, contextually relevant text based on input prompts. This includes content creation, summarization, and language translation. Its code generation capabilities assist developers by generating programming code, debugging, and offering coding suggestions, enhancing productivity. The service’s image generation capabilities allow for the creation of detailed, customized visuals from textual descriptions, supporting creative and design processes across various industries by transforming words into images.

Exam Prep

Preparing for the AI-900 exam involved a structured approach that began with reviewing the official Microsoft exam guide to understand the key topics and skills measured. I allocated specific time blocks for each topic, starting with foundational concepts of AI and machine learning, progressing through Azure’s specific tools and services like Azure Cognitive Services, Azure Machine Learning, and the principles of responsible AI.

I supplemented my study with online courses and tutorials specifically designed for the AI-900 exam, focusing on both theoretical concepts and practical applications within the Azure ecosystem.

Exam simulations played a significant role in my preparation. I regularly tested myself with practice exams that mimicked the format and difficulty of the real test. These simulations helped me identify areas of weakness, allowing me to dedicate more study time to those areas. They also familiarized me with the exam’s time constraints and question formats, improving my time management during the actual exam.

Conclusion

Reflecting on the experience, this methodical and comprehensive approach, combined with the consistent use of exam simulations, was instrumental in passing the AI-900 exam. It not only prepared me for the exam itself but also provided a solid foundation in AI and machine learning principles, particularly within the Azure platform.

--

--