Advanced Level

AWS Certified Machine Learning Specialty MLS-C01 2026

Master machine learning theory and SageMaker tooling to design and deploy ML solutions at scale — and pass the demanding MLS-C01 Specialty exam.

★ 4.8 (300 learners) 60 hrs total 5 modules 3 capstones

AWS EC2 S3 IAM VPC SageMaker +7 more

Only 8 seats left · Next batch starts 5 Sept 2026

Download Curriculum

60 hrs: of content
5 modules: in curriculum
19 projects: hands-on builds
Advanced: difficulty

Curriculum covers

Frameworks you'll master

13 tools & frameworks

AWS EC2 S3 IAM VPC SageMaker Amazon Bedrock Apache Spark Kubernetes ChatGPT Python AWS Glue Step Functions

Course Curriculum

5 modules · 60 hours · 24 live sessions

Data Engineering 6 sessions

Section Intro: Data Engineering
[Important] AWS Console UI Update
Set up an AWS Billing Alarm
Amazon S3 – Overview
Amazon S3 Storage Classes + Glacier
Amazon S3 Storage + Glacier Hands On
Amazon S3 Lifecycle Rules (with S3 Analytics
Amazon S3 Lifecycle Rules Hands On)
Amazon S3 Bucket Policy
Amazon S3 – Encryption
Amazon S3 Default Encryption
Amazon S3 VPC Endpoints
Amazon FSx
Amazon Kinesis Data Streams
Amazon Data Firehose
Amazon Data Firehose
Amazon Managed Service for Apache Flink
Lab 1.2 Managed Service for Apache Flink (Formerly Kinesis Data Analytics)
Kinesis Video Streams
Kinesis ML Summary
Glue Data Catalog & Crawlers
Lab 1.3 Glue Data Catalog
Glue ETL
Lab 1.4 Glue ETL
Glue Data Brew
Lab 1.5 Glue Data Brew
Lab 1.6 – Athena
AWS Data Stores in Machine Learning
AWS Data Pipelines
AWS Batch
AWS DMS Database Migration Services
AWS Step Functions
Full Data Engineering Pipelines
Random things you need to know: AWS DataSync and MQTT
Data Engineering Summary

Exploratory Data Analysis 3 sessions

Section Intro: Data Analysis
Python in Data Science and Machine Learning
Example: Preparing Data for Machine Learning in a Jupyter Notebook.
Types of Data
Data Distributions
Time Series: Trends and Seasonality
Introduction to Amazon Athena
Overview of Amazon Quicksight
Types of Visualizations, and When to Use Them.
Elastic MapReduce (EMR) and Hadoop Overview
Apache Spark on EMR
EMR Notebooks, Security, and Instance Types
Feature Engineering and the Curse of Dimensionality
Imputing Missing Data
Dealing with Unbalanced Data
Handling Outliers
Binning, Transforming, Encoding, Scaling, and Shuffling
Amazon SageMaker Ground Truth and Label Generation
Lab: Preparing Data for TFIDF with Spark and EMR Studio, Part 1
Lab: Preparing Data for TFIDF with Spark and EMR Studio, Part 2

Modelling, Part 1: General Deep Learning and Machine Learning 11 sessions

Section Intro: Modelling
Introduction to Deep Learning
Activation Functions
Convolutional Neural Networks
Recurrent Neural Networks
Modern NLP with BERT and GPT, and Transfer Learning
Tuning Neural Networks
Regularization Techniques for Neural Networks (Dropout, Early Stopping)
Deep Learning on EC2 and EMR
L1 and L2 Regularization
Grief with Gradients: The Vanishing Gradient problem
The Confusion Matrix
Precision, Recall, F1, AUC, and more
Ensemble Methods: Bagging and Boosting
Modelling, Part 2: Amazon Sagemaker
Introducing Amazon SageMaker
UPDATE: SageMaker is now "SageMaker AI"
Linear Learner in SageMaker
XGBoost in SageMaker
Seq2Seq in SageMaker
DeepAR in SageMaker
BlazingText in SageMaker
Object2Vec in SageMaker
Object Detection in SageMaker
Image Classification in SageMaker
Semantic Segmentation in SageMaker
Random Cut Forest in SageMaker
Neural Topic Model in SageMaker
Latent Dirichlet Allocation (LDA) in SageMaker
KNearestNeighbours (KNN) in SageMaker
KMeans Clustering in SageMaker
Principal Component Analysis (PCA) in SageMaker
Factorization Machines in SageMaker
IP Insights in SageMaker
Reinforcement Learning in SageMaker
Automatic Model Tuning
Apache Spark with SageMaker
SageMaker Studio, and SageMaker Experiments
SageMaker Debugger
SageMaker Autopilot / AutoML
SageMaker Model Monitor
Deployment Guardrails and Shadow Tests
WARNING about SageMaker billing
SageMaker Canvas
Bias Measures in SageMaker Clarify
SageMaker Training Compiler
SageMaker Feature Store
SageMaker ML Lineage Tracking
SageMaker Data Wrangler
Demo: SageMaker Studio, Canvas, and Data Wrangler
Modelling, Part 3: HighLevel ML Services
Amazon Comprehend
Amazon Translate
Amazon Transcribe
Amazon Polly
Amazon Rekognition
Amazon Forecast
Amazon Forecast Algorithms
Amazon Lex
Amazon Personalize
Lightning round! TexTract, DeepRacer, Lookout, and Monitron
TorchServe and AWS Neuron
Deep Composer, Fraud Detection, CodeGuru, and Contact Lens
Amazon Kendra and Amazon Augmented AI (A2I)
Modelling, Part 4: Wrapping up & Lab Activity
Putting them All Together
Lab: Tuning a Convolutional Neural Network on EC2, Part 1
Lab: Tuning a Convolutional Neural Network on EC2, Part 2
Lab: Tuning a Convolutional Neural Network on EC2, Part 3

ML Implementation and Operations 2 sessions

Section Intro: Machine Learning Implementation and Operations
SageMaker's Inner Details and Production Variants
SageMaker On the Edge: SageMaker Neo and IoT Greengrass
SageMaker Security: Encryption at Rest and In Transit
SageMaker Security: VPC's, IAM, Logging, and Monitoring
SageMaker Resource Management: Instance Types and Spot Training
SageMaker Resource Management: Automatic Scaling, AZ's
SageMaker Serverless Inference and Inference Recommender
SageMaker Inference Pipelines
MLOps with SageMaker, Kubernetes, SageMaker Projects, and SageMaker Pipelines
Lab: Tuning, Deploying, and Predicting with TensorFlow on SageMaker Part 1
Lab: Tuning, Deploying, and Predicting with TensorFlow on SageMaker Part 2
Lab: Tuning, Deploying, and Predicting with TensorFlow on SageMaker Part 3

Generative AI: Transformers, GPT, SelfAttention and Foundation Models 2 sessions

The Transformer Architecture
SelfAttention and Attentionbased Neural Networks in Depth
Applications of Transformers (such as GPT4 and ChatGPT)
Generative PreTrained Transformers (GPT) in Depth: How They Work, Part 1
Generative PreTrained Transformers (GPT) in Depth: How They Work, Part 2
FineTuning / Transfer Learning with GPT and Transformers
Lab: Tokenization and Positional Encoding with SageMaker Notebooks + Hugging face
Lab: MultiHeaded, Masked SelfAttention in Sagemaker and Hugging face
Lab: Using GPT within a SageMaker Notebook
AWS Foundation Models and Amazon SageMaker JumpStart with Generative AI
Lab: Using Amazon SageMaker JumpStart to load and use GPT from Hugging Face
Building Generative AI with Amazon Bedrock and Foundation Models
Lab: Chat, Text, and Image Foundation Models in the Bedrock Playground
Amazon Q Developer (formerly Codewhisperer)
AWS Health Scribe AIgenerated clinical notes from consultation transcriptions

Download Full Curriculum

After this course, you will

Skills you carry to every job

Select the correct ML algorithm (XGBoost, CNN, LSTM, K-Means) for each business problem
Engineer features, handle class imbalance, and prepare tabular, text, and image datasets for training
Train, tune, and evaluate models using SageMaker training jobs and Hyperparameter Optimization
Deploy models to real-time, batch, and asynchronous inference endpoints with auto-scaling
Detect and mitigate bias using SageMaker Clarify and implement model explainability
Build generative AI solutions using Amazon Bedrock foundation models and RAG architectures
Implement MLOps pipelines with SageMaker Pipelines, Model Registry, and automated deployment
Pass the AWS Certified Machine Learning Specialty MLS-C01 exam

Hands-on learning

What you'll build

3 industry capstone projects · built from scratch · yours to keep

Fraud Detection Model at Scale

Build a binary classification fraud detection system: ingest transaction data from S3, perform feature engineering with SageMaker Data Wrangler, train an XGBoost model with SageMaker HPO, evaluate with Clarify for bias, deploy to a real-time endpoint, and set up Model Monitor for data drift alerts. Process 1M+ transactions with batch transform.

SageMaker SageMaker Clarify SageMaker Model Monitor S3 AWS Glue Athena CloudWatch Lambda

Computer Vision Product Defect Classifier

Train a CNN image classifier using SageMaker with a custom PyTorch container on GPU instances. Use SageMaker Ground Truth to label a defect image dataset, apply data augmentation, evaluate precision/recall, and deploy with SageMaker multi-model endpoints for cost efficiency. Integrate with an IoT Greengrass edge device for real-time inference.

SageMaker SageMaker Ground Truth ECR S3 Python PyTorch CloudWatch IoT Greengrass

Generative AI Text Summarisation Service

Fine-tune a foundation model on Amazon Bedrock using a domain-specific dataset, implement a RAG pipeline with Bedrock Knowledge Bases and an OpenSearch vector store, expose via API Gateway with Lambda, and add guardrails for content filtering. Evaluate output quality using ROUGE and BLEU scores.

Amazon Bedrock SageMaker OpenSearch Lambda API Gateway S3 CloudWatch Python

Plus 16 guided mini-projects across the curriculum — Q&A bots, resume screeners, SQL helpers, and more.

Batch Schedule

⚡ Seats filling fast — only 8 remaining across all batches

Available Batches
Batch Type	Schedule	Duration	Next Start Date	Availability
Weekend	Sat & Sun, 10am–1pm	10 weeks	5 Sept 2026	Available
Weekday	Mon–Fri, 7–9pm	8 weeks	7 Sept 2026	Available
Fast-track	Mon–Sun, 6 hours/day	2 weeks	14 Sept 2026	Available

Frequently Asked Questions

How much mathematics is needed for the MLS-C01 exam?

The exam tests practical algorithm selection rather than theorem proofs. You should understand gradient descent, overfitting vs underfitting, precision vs recall, L1/L2 regularisation, and basic linear algebra at a conceptual level.

What programming languages and frameworks are covered?

Python is the primary language. The course covers scikit-learn, PyTorch, TensorFlow/Keras, Spark MLlib on EMR, and SageMaker's built-in algorithms. Framework-specific deep dives are provided for the most exam-relevant ones.

How does the ML Specialty differ from the ML Engineer Associate?

The MLS-C01 Specialty is broader and deeper — it covers ML theory, algorithm selection, and advanced SageMaker. The MLA-C01 Associate focuses on MLOps and deploying models operationally. MLS-C01 is the more demanding exam.

How many practice questions are included?

Two full 65-question practice exams are included, plus topic-level quizzes. Detailed answer explanations cover why each wrong answer is incorrect, which is critical for Specialty-level preparation.

Is generative AI part of the MLS-C01 exam?

Yes. Amazon Bedrock, foundation model fine-tuning, prompt engineering, RAG architectures, and responsible AI for generative models have been added to the updated MLS-C01 exam blueprint.

What salary premium does the ML Specialty typically offer?

AWS ML Specialty certified professionals command salaries 20–35% higher than uncertified peers in data science and ML engineering roles, according to Global Knowledge and Dice salary surveys.

Ready to Get Started?