How I prepared and passed the exam.
In November 2020, I decided to pursue cloud certification with not much prior knowledge about it. Through research on industry demands, I found AWS Certified Machine Learning — Specialty (MLS-C01) to be the closest on my path at the time. I passed the exam with less than 3 months of preparation. In this post, I will provide you an overview of resources and strategies I used for passing the exam. I hope it will serve you well on your certification journey.
MLS-C01 is one of the advanced level certifications that AWS offers, and it is intended for individuals who perform a development or data science role. According to AWS, it validates an examinee’s ability to build, train, tune, and deploy acMachine Learning (ML) models using the AWS cloud, specifically:
- Select and justify the appropriate ML approach for a given business problem.
- Identify appropriate AWS services to implement ML solutions.
- Design and implement scalable, cost-optimized, reliable, and secure ML solutions.
AWS recommends having 1–2 years of hands-on experience developing, architecting, or running ML/Deep Learning (DL) workloads on the AWS cloud. Although I had experience building ML/DL models on local machine with tools such as R and Python, I did not have their recommended experiences on the cloud. My most hands-on experience with on AWS occurred during exam preparation.
One of the challenging aspects of this exam is collecting study materials. There are numerous online resources and materials you can easily find. Sometimes you get drowned in the resources without progress. The key is to prepare strategically with target exam date in sight, cover important topics and go for it. You can’t prepare enough for this exam; study materials can be unlimited. Trying to get a perfect score is not the best approach. This is a pass-fail exam. Score is confidential to you. If you know the guidelines and topics you have chosen with enough practice; you should feel good about it.
Exam contains 65 scenario-based multiple choice and multiple response questions. Unanswered questions are scored as incorrect; there is no penalty for guessing. Your results for the examination are reported as a score from 100–1,000, with a minimum passing score of 750. You will know pass/fail message immediately after submission (for online) and receive the official badge and certificate within 5 days.
You will be tested on domains of Data Engineering, Exploratory Data Analysis (EDA), Machine Learning and Deep Learning using AWS services. Unlike other AWS certifications, in addition to AWS services, this exam also focuses on non-AWS aspects i.e. ML principle and EDA. The list below are the main content domains and their weightings for the exam. It is not a comprehensive listing of the content and service.
Domain 1: Data Engineering (20%)
This domain deals with building and implementing data repositories, data-ingestion and data-transformation solution. You are expected to have knowledge of what these services do and when to use them.
S3, Kinesis, Glue, RDS (Aurora), NoSQL(DynamoDB), Batch, Data Pipeline, Data Migration Service(DMS), Lambda, Step Functions, Redshift, Lake Formation, Athena, QuickSight, EMR.
Domain 2: Exploratory Data Analysis (24%)
This domain is mostly non-AWS related. If you have taken statistics classes or done data analysis works you are in good shape. It involves feature engineering techniques such as one-hot encoding, normalization, standardization, handling missing values, transformation, understanding basis visualization techniques, distribution or detecting outliers. Additionally, you may be asked question using visualization service such as Athena, QuickSight and big data processing tool like EMR in combination with other services.
Domain 3: Modeling (36%)
Modeling is the most significant section of the exam. It considers framing business problems as ML problems, selecting appropriate algorithms, training models, hyperparameter tuning and evaluating models; all revolves around Amazon’s signature service SageMaker with ML/DL concepts. I encourage you to read and understand SageMaker‘s developer guide including architecture, built-in algorithms and their use cases. Topics related to this domain are as follows.
SageMaker, ML/DL Algorithms and Concepts, Deep Learning Frameworks, Neural Networks, Activation Functions, Hyperparameter Tuning, Regularization Techniques, Confusion Matrix.
Domain 4: Machine Learning Implementation and Operations (20%)
This section consists of building and implementing ML solutions for performance, availability, scalability, resiliency, and fault tolerance, security and monitoring, deploying in production and understanding AI services.
Below are the topics related to this domain.
Automatic Scaling, Instant Types, Batch Transformation, Incremental and Spot Training, Inference (Elastic Inference, Inference Pipeline), SageMaker Neo, Edge Devices, Model Monitor(CloudWatch), High level AI Services, ECR, Availability Zone (AZ), Security (Encryption, VPC, IAM, CloudTrail).
1. Udemy Course
The first step I took for the journey was Udemy course, AWS Certified Machine Learning Specialty 2021 — Hands On. This course provides a high-level overview of exam and study materials. This is a great course to get started. I watched it three times. First, I just watched through the course to feel for the exam. Most of the data engineering section was new for me although I did have some experience in building data pipelines. Second, I watched lectures with notes, referenced material and hands-on practice. Third, I revisited the course a week before the exam — by that time I had taken practice exams, studied AWS documentations and other referenced materials. I felt good about content and exam. This course was just a beginning.
2. AWS Documentations
If there is one resource I have to recommend, it would be AWS’s own materials including documentations, product overview and it’s use case, FAQs, E-Learning courses. I highly recommend AWS Cloud Practitioner Essentials (CPE) course to those who are new to the cloud concepts and AWS Services. It provides overview of AWS cloud, independent of specific technical roles. You won’t be quizzed on the CPE but AWS cloud knowledge on topics such as security, architecture is useful in understanding services and system you will be asked in the exam. AWS blogs are additional sources if you want to understand specific algorithms or service use cases in details.
3. Practice Exams
Practice exams are important. First, it gives you an idea where you stand in terms of preparation. It doesn’t mean you will do better or worse, but it’s a good indicator. Secondly, it encourages you to think strategically from exam perspective while studying; typically, you just try to understand or remember. I found this about a lot when I took first practice test, which I did poorly. I thought I knew a lot about materials. I was wrong, mainly due to not having enough practices in exam scenario. Having expectations of how questions are asked adds another viewpoint to the study. I practiced Udemy exam courses. Additionally, I tried every online practice test including AWS course quiz.
Most questions are based on use case scenario where you will be asked a question on a real business problem. Some questions are a paragraph long, some even longer. Don’t be intimated by the length of questions. Try to understand the questions by making mental note of key phrases and qualifiers such as efficient, cost effective, no-maintenance and come up with possible answers before looking at the options. You are not allowed to take physical notes. Use technique of elimination to reduce answer choices. You should be able to eliminate one or two options without much effort. Some options will not make sense, they are there just as a distraction. Elimination option is even more useful when you encounter questions from the topic you are not familiar with. If you can eliminate one or two options, you increase chance of getting answer right from 25% to 33% or 50%. If you plan for other AWS exams like AWS Certified Cloud Practitioner, take those first. In addition to providing foundation to the ML Specialty exam, you get 50% discount on the next exam.
You can go to a link AWS Certification and create account to schedule for exam. Pick eligible exams from the list, PSI or Pearson VUE. Exam dates are readily available. I chose Pearson VUE online version. You can check-in 30 minutes before start time. It should be fairly easy process given you have read instructions and guidelines prior. You could take exam once check-in is complete. Your time starts moment you see first question on the screen. It’s a three-hour exam. Remember, you are not allowed to take notes or use calculator. During exam, you can flag or skip questions for later review. Take advantages of those options. Don’t spend too much trying to answer hard question. Flag it and comeback later. Stamina is key, take all necessary steps to be productive.
All in all, I found this certification exam to be challenging but certainly doable with right strategy, study material and practice. I hope this post has been helpful in your preparation for MLS-C01.Thank you for reading. I wish you all the best on your AWS Cloud Certification Journey!
2. Exam Readiness: AWS Machine Learning Specialty
3. AWS Training and Certification Question
4. AWS Certified Machine Learning Specialty Full Practice Exam
5. AWS Certified Machine Learning Specialty: 3 PRACTICE EXAMS