Amazon SageMaker & Machine Learning in the Cloud
Goal - To test your understanding of Machine Learning and related Analytics use cases;
In-depth challenge of your Sagemaker platform knowledge (as a whole);
how it interacts with other parts of AWS; and
Related Data Preparation & Analytics;
your know-how in the ML space (this is AWS agnostic)
This course can strengthen your foundations for both the ML as well as the Data Analytics exams with its use case questions.
Purpose - The purpose of this is to impart in-depth knowledge of Amazon Sagemaker and applied ML to students via the most powerful teaching medium: questions that test your level of understanding. While this should definitely help them do well in AWS ML certification (the hardest exam!), knowing the ins-and-outs of Sagemaker and Applied ML is expected to make their professional lives (in the ML space) successful. This course will challenge the knowledge of even the experienced practitioner primarily to bring out all the nuances of Applied ML.
The course is organized via three tests that are designed to bring out deep nuances associated with the setup, development, working, algorithmic, and operational aspects of ML know-how, Sagemaker, and How-to-dos.
Details - The three tests (50 questions each) cover the following aspects:
Machine Learning - Core Foundational Knowledge (this is independent of AWS and Sagemaker).
AI Services in AWS - AI application service knowledge including but not limited to Lex, Transcribe, Translate, Comprehend, Rekognition, ...
ML Development Lifecycle Management & Administration- Sagemaker domains, Studio, Notebooks, ML Environments, CRISP/DM concepts.
Labeling & Ground Truth
Process Data - Data life cycle in Sagemaker, AWS services involved in data life cycle, Data wrangler. Batch data processing, and Data processing integration with Bigdata environments (EMR). Data visualization. How data services in AWS can integrate in.
Training - Sagemaker choice of algorithms (built-in), Setup of Training, Training Jobs and details, Distributed Training. Sagemaker and its interaction with other services for getting training data. Sources of training data and nuances. Built in algorithm capabilities. Modeling choices and what best fits.
Inference & its life cycle - ML Ops in as well as outside the Cloud.
Model debugging and monitoring - Model monitoring and debugging minutiae.
Security - Data as well as Infrastructure
ML Data Analytics - Data Analytics eco system in AWS for ML and relevant aspects of other pertinent AWS services incl. Athena, Glue Kinesis, and EMR emphasis (also helps Data Analytics foundations)
For MLS-C01 exam takers
Broad guidelines based on my experience below. The curriculum is so vast, so any single "practice test" course is unlikely to meet all of it. Hence. questions are designed so the test will cover all the four domains via applied problem-solving questions and approximately weighted by curriculum spec to maximize its impact: i.e., Modeling takes highest precedence, then exploratory data analysis, followed by data engineering and MLOps (both approximately equal weighted).
Machine Learning Exam curriculum in AWS is specified as four domains, but it is actually 3 major sub-paths:
ML and Data/Feature Engineering - Analytic thinking to analyze/decompose problem, data-process/standardize/clean and map to the most apt solution. This has pretty much nothing to do with AWS at all, but you should know the components of Sagemaker that helps here.
Sagemaker based ML Data Modeling + Surrounding Ecosystem: This is a major part of the exam, but a deeper knowledge here goes beyond the exam.
AWS overall architecture emphasizing security, encryption, storage, compute - I suggest that you clear an associate level exam in AWS to be honest to handle this set of questions easily (about 10-15%). Plenty of resource materials including good courses/exams in Udemy, and elsewhere are available. Of course, we will cover the most critical and relevant aspects in the context of points in #1 and #2 above.
The AWS exam is 65 questions but only 50 is used for assessments.
Useful Exam Prep Resource: Please review AWS Skill Builder for free publicly available AWS question sets (incl. practice exams in Skill Builder) to further "Data Augment" the knowledge provided by these tests.
Focus on the pattern of ask in the practice test questions, not just the question.
Who this course is for:
- Persons trying to develop indepth and diagnostic knowledge of Sagemaker; Persons preparing for AWS ML Certification
- Data scientists, Engineers, and ML enthusiasts
- For those who curious about data analysis basics
Gokul Prabhakar has a Ph.D. in Computer Science specializing in AI (Knowledge Based Systems, and Multi-agents). He worked at Bell Laboratories (Research) for more than 7 years in the areas of Expert Systems for automated diagnosis, and Intelligent Network Management. Gokul Prabhakar has several publications in leading conferences and journals which can be viewed via Google Scholar, and holds US Patents for four inventions. Gokul Prabhakar also serves in the Editorial Boards of some leading journals in the area of Data Analysis and Applied Intelligence. He was also invited as a Panelist for IEEE ANTS ‘Ubiquitous Digital Connectivity and Economic Prosperity for All – Financial Inclusion with FinTech’ (Dec 2021). He recently completed the AWS Arch. Associate AWS SAA-C02 (Aug 2021), AWS Machine Learning Specialty MLS-C01 (Jan 2022), and AWS Data Analytics Specialty DAS-C01 (Apr 2022) certifications, and developed an interest to form courses in Udemy to help students trying to further knowledge in the area of AI/ML. Data Analytics, and related aspects for certifications as well as for sharpening their deep insights in this area.