Free MLS-C01 Practice Test Questions and Answers (2026)

Last Update Check

View Mode
Q: 1
[Modeling] An agricultural company is interested in using machine learning to detect specific types of weeds in a 100-acre grassland field. Currently, the company uses tractor-mounted cameras to capture multiple images of the field as 10 × 10 grids. The company also has a large training dataset that consists of annotated images of popular weed classes like broadleaf and non-broadleaf docks. The company wants to build a weed detection model that will detect specific types of weeds and the location of each type within the field. Once the model is ready, it will be hosted on Amazon SageMaker endpoints. The model will perform real-time inferencing using the images captured by the cameras. Which approach should a Machine Learning Specialist take to obtain accurate predictions?
Options
18 comments in the community discussion
2
C
2
Yeah, gotta go with C. SageMaker SSD requires RecordIO for object detection, and image classification (A or D) won't return weed locations, which is needed here. Parquet format (B, D) is a distractor. Pretty sure about this but let me know if I'm off.
Q: 2
[Data Engineering] A retail company is ingesting purchasing records from its network of 20,000 stores to Amazon S3 by using Amazon Kinesis Data Firehose. The company uses a small, server-based application in each store to send the data to AWS over the internet. The company uses this data to train a machine learning model that is retrained each day. The company's data science team has identified existing attributes on these records that could be combined to create an improved model. Which change will create the required transformed records with the LEAST operational overhead?
Options
22 comments in the community discussion
4
Going with A here too. Using Lambda for transformation within Firehose avoids managing infrastructure, and AWS handles the scaling. The other choices involve running clusters or EC2, which is more to maintain. Pretty sure this is the least ops work, correct me if I missed something!
3
A . Using Lambda for transformation with Firehose means no server management, auto-scaling, and it's built right into the delivery stream. The other options add way more operational work. If everything fits in a Lambda, this is the easiest path. Tell me if I'm missing any curveballs here.
Q: 3
[Data Engineering] A large JSON dataset for a project has been uploaded to a private Amazon S3 bucket The Machine Learning Specialist wants to securely access and explore the data from an Amazon SageMaker notebook instance A new VPC was created and assigned to the Specialist How can the privacy and integrity of the data stored in Amazon S3 be maintained while granting access to the Specialist for analysis?
Options
23 comments in the community discussion
4
C . ACLs in A are risky and presigned URLs in D don’t meet the privacy/integrity part. B is a trap since copying local doesn’t control S3 access. Saw a similar question in some practice-C is the secure setup with VPC endpoint plus bucket policy. Correct me if I missed something on endpoint restrictions.
3
C. had something like this in a mock and C was the answer. VPC endpoint plus custom bucket policy limits access securely.
Q: 4
[Data Engineering] A medical imaging company wants to train a computer vision model to detect areas of concern on patients' CT scans. The company has a large collection of unlabeled CT scans that are linked to each patient and stored in an Amazon S3 bucket. The scans must be accessible to authorized users only. A machine learning engineer needs to build a labeling pipeline. Which set of steps should the engineer take to build the labeling pipeline with the LEAST effort?
Options
21 comments in the community discussion
6
C . Private workforce in SageMaker Ground Truth covers "authorized users only" since you control access, and using the built-in bounding box task means no custom labeling app needed. That keeps effort low compared to building your own tools or using Mechanical Turk (which isn't private enough for this kind of medica
4
C . You get private access control with SageMaker Ground Truth, and the bounding box task is built-in so no custom tool needed. Seems quickest for sensitive healthcare data, though open to other takes if I missed something.
Q: 5
[Data Engineering] A machine learning specialist is preparing data for training on Amazon SageMaker. The specialist is using one of the SageMaker built-in algorithms for the training. The dataset is stored in .CSV format and is transformed into a numpy.array, which appears to be negatively affecting the speed of the training. What should the specialist do to optimize the data for training on SageMaker?
Options
21 comments in the community discussion
6
Option C, Had something like this in a mock, and SageMaker built-in algorithms are definitely optimized for RecordIO protobuf format. Using numpy arrays can slow down training since the built-ins expect RecordIO for efficiency. Not 100% but pretty sure C is the way to go here, agree?
Its C here. The built-in SageMaker algorithms are optimized for RecordIO protobuf, not numpy arrays or even Parquet, which is mostly about storage efficiency. D feels like a trap-hyperparameter optimization only tunes model params, not the raw data format. Pretty sure C, but if I missed something let me know.
Q: 6
[Data Engineering] A company wants to predict stock market price trends. The company stores stock market data each business day in Amazon S3 in Apache Parquet format. The company stores 20 GB of data each day for each stock code. A data engineer must use Apache Spark to perform batch preprocessing data transformations quickly so the company can complete prediction jobs before the stock market opens the next day. The company plans to track more stock market codes and needs a way to scale the preprocessing data transformations. Which AWS service or feature will meet these requirements with the LEAST development effort over time?
Options
13 comments in the community discussion
1
B
1
A tbh
Q: 7
[Modeling] A beauty supply store wants to understand some characteristics of visitors to the store. The store has security video recordings from the past several years. The store wants to generate a report of hourly visitors from the recordings. The report should group visitors by hair style and hair color. Which solution will meet these requirements with the LEAST amount of effort?
Options
18 comments in the community discussion
1
C or A. If we strictly care about the least effort, semantic segmentation (C) should work best for isolating hair regions, then ResNet-50 is purpose-built for classes like hairstyle and color. But object detection (A) could get close if the hair region is obvious in most frames. I think C has the edge, but in a dataset
1
For me, C, semantic segmentation plus ResNet-50 lines up with least effort for visual grouping like this.
Q: 8
[Data Engineering] A Machine Learning Specialist is developing a daily ETL workflow containing multiple ETL jobs The workflow consists of the following processes * Start the workflow as soon as data is uploaded to Amazon S3 * When all the datasets are available in Amazon S3, start an ETL job to join the uploaded datasets with multiple terabyte-sized datasets already stored in Amazon S3 * Store the results of joining datasets in Amazon S3 * If one of the jobs fails, send a notification to the Administrator Which configuration will meet these requirements?
Options
20 comments in the community discussion
9
Option A looks right to me. Step Functions can coordinate the ETL workflow and wait for uploads, then Glue is ideal for joining large datasets in S3. CloudWatch with SNS covers notifications. Pretty sure this matches the scenario, unless I’m missing something.
1
Makes sense to pick A here. Step Functions can coordinate all dataset arrival events before kicking off Glue, which is ideal for large data joins, plus CloudWatch + SNS covers the failure notification part. Pretty sure that's the setup they want, unless I'm missing something.
Q: 9
[Modeling] While working on a neural network project, a Machine Learning Specialist discovers thai some features in the data have very high magnitude resulting in this data being weighted more in the cost function What should the Specialist do to ensure better convergence during backpropagation?
Options
19 comments in the community discussion
5
Option B makes sense here. When features have very different scales, normalization is key since it evens out their influence on the cost function and helps gradients flow better during backprop. Regularization (C) helps with overfitting but doesn't fix scaling problems directly. I'm confident about B, but open if so
1
I don't think it's A. Dimensionality reduction drops features but doesn't solve the scaling issue. I usually see B (data normalization) in practice for this, since it keeps big features from overpowering the others during backprop. C is tempting but more about overfitting. Anyone see it different?
Q: 10
[Data Engineering] A company has raw user and transaction data stored in AmazonS3 a MySQL database, and Amazon RedShift A Data Scientist needs to perform an analysis by joining the three datasets from Amazon S3, MySQL, and Amazon RedShift, and then calculating the average-of a few selected columns from the joined data Which AWS service should the Data Scientist use?
Options
18 comments in the community discussion
2
A . Athena supports federated queries with connectors for S3, MySQL, and Redshift, so it can join across all three sources in a single query. B is tempting but Spectrum can’t directly query MySQL. Correct me if I missed something though!
2
Option A is what I'd pick here. Pretty sure Athena can query directly from S3, and it can also use federated queries to pull data from MySQL and Redshift. Not 100% because I get confused with Glue sometimes, but Athena seems best for actually joining across these sources and running SQL-type analysis. Anyone else ag
Q: 11
[Exploratory Data Analysis] A machine learning (ML) engineer is preparing a dataset for a classification model. The ML engineer notices that some continuous numeric features have a significantly greater value than most other features. A business expert explains that the features are independently informative and that the dataset is representative of the target distribution. After training, the model's inferences accuracy is lower than expected. Which preprocessing technique will result in the GREATEST increase of the model's inference accuracy?
Options
3 comments in the community discussion
Pretty sure it's A. Normalizing those features keeps their info but gets rid of scale problems. Makes sense in this scenario.
Q: 12
[Modeling] A finance company needs to forecast the price of a commodity. The company has compiled a dataset of historical daily prices. A data scientist must train various forecasting models on 80% of the dataset and must validate the efficacy of those models on the remaining 20% of the dataset. What should the data scientist split the dataset into a training dataset and a validation dataset to compare model performance?
Options
4 comments in the community discussion
Had something like this in a mock, it's A for time series so you don't leak future info into training.
Actually, for time series data you shouldn't use random sampling like D says. A is the way to go.
Q: 13
[Modeling] A company that runs an online library is implementing a chatbot using Amazon Lex to provide book recommendations based on category. This intent is fulfilled by an AWS Lambda function that queries an Amazon DynamoDB table for a list of book titles, given a particular category. For testing, there are only three categories implemented as the custom slot types: "comedy," "adventure,” and "documentary.” A machine learning (ML) specialist notices that sometimes the request cannot be fulfilled because Amazon Lex cannot understand the category spoken by users with utterances such as "funny," "fun," and "humor." The ML specialist needs to fix the problem without changing the Lambda code or data in DynamoDB. How should the ML specialist fix the problem?
Options
4 comments in the community discussion
1
D makes sense, but what if users say something unexpected that's not a direct synonym? Sometimes Lex can't generalize unless synonyms are mapped specifically. In similar exam questions, AWS is picky about using synonyms over creating extra slot types or adding to enumeration values. I think D is right, but it really
D imo. Adding those words as synonyms in the custom slot type lets Lex map "funny", "humor" etc to "comedy" without touching code or DynamoDB. Pretty confident that's what AWS recommends for these cases.
Q: 14
[Modeling] A company that manufactures mobile devices wants to determine and calibrate the appropriate sales price for its devices. The company is collecting the relevant data and is determining data features that it can use to train machine learning (ML) models. There are more than 1,000 features, and the company wants to determine the primary features that contribute to the sales price. Which techniques should the company use for feature selection? (Choose three.)
Options
5 comments in the community discussion
1
B D, and E are right if they're specifically asking about feature selection techniques. But if the question meant features to help normalize or preprocess data instead of just select important ones, then A might be in play too. Does the question require picking the most useful for selection specifically
Q: 15
[Machine Learning Implementation and Operations] A Machine Learning Specialist deployed a model that provides product recommendations on a company's website Initially, the model was performing very well and resulted in customers buying more products on average However within the past few months the Specialist has noticed that the effect of product recommendations has diminished and customers are starting to return to their original habits of spending less The Specialist is unsure of what happened, as the model has not changed from its initial deployment over a year ago Which method should the Specialist try to improve model performance?
Options
3 comments in the community discussion
1
Option D is the way to go. Retraining with original data plus new data is key since user behavior and inventory shift over time. B is a trap here, just tuning hyperparams won't fix stale training data. Seen similar on other practice sets.
1
D imo. Model drift is likely, so adding new data as inventory changes makes sense.
Q: 16
[Data Engineering] A data scientist is using an Amazon SageMaker notebook instance and needs to securely access data stored in a specific Amazon S3 bucket. How should the data scientist accomplish this?
Options
5 comments in the community discussion
1
I get the point about A, but I'm thinking C is correct for secure access with SageMaker roles.
1
Maybe A. Granting permissions using a bucket policy directly seems like it covers access for the SageMaker notebook, right?
Q: 17
[Modeling] An online reseller has a large, multi-column dataset with one column missing 30% of its data A Machine Learning Specialist believes that certain columns in the dataset could be used to reconstruct the missing data. Which reconstruction approach should the Specialist use to preserve the integrity of the dataset?
Options
5 comments in the community discussion
1
I don’t think it’s B. Multiple imputation (C) is more robust here since it uses other columns to estimate missing values, which helps maintain statistical integrity. Last observation carried forward works best for time series but not general datasets like this.
Seen similar on the official practice test, pretty sure it's C.
Q: 18
[Machine Learning Implementation and Operations] A bank's Machine Learning team is developing an approach for credit card fraud detection The company has a large dataset of historical data labeled as fraudulent The goal is to build a model to take the information from new transactions and predict whether each transaction is fraudulent or not Which built-in Amazon SageMaker machine learning algorithm should be used for modeling this problem?
Options
7 comments in the community discussion
1
B . Had something like this in a mock, labeled fraud data fits XGBoost.
D . Random Cut Forest is for anomaly detection and fraud is basically an anomaly, right? Whenever I see "fraud detection" it's easy to fall for the unsupervised option. Unless those labels are strictly needed by the question, I could see D being picked. Someone double check me on that.
Q: 19
[Modeling] A Machine Learning Specialist is building a model to predict future employment rates based on a wide range of economic factors While exploring the data, the Specialist notices that the magnitude of the input features vary greatly The Specialist does not want variables with a larger magnitude to dominate the model What should the Specialist do to prepare the data for model training'?
Options
9 comments in the community discussion
5
My pick: C Normalization is how you prevent features with large values from affecting the model more than others.
1
C not A. Normalization handles the magnitude issue best here, don't think quantile binning (A) does what the question asks.
Q: 20
[Data Engineering] A company's machine learning (ML) specialist is building a computer vision model to classify 10 different traffic signs. The company has stored 100 images of each class in Amazon S3, and the company has another 10.000 unlabeled images. All the images come from dash cameras and are a size of 224 pixels * 224 pixels. After several training runs, the model is overfitting on the training data. Which actions should the ML specialist take to address this problem? (Select TWO.)
Options
5 comments in the community discussion
1
C/E? Had something like this in a mock. Data augmentation (C) adds variety so the model generalizes better, and semi-supervised with k-NN (E) helps expand the labeled set without manual effort. Not 100 percent certain but this matches most practice explanations, agree?
1
Probably C and E. Data augmentation (C) increases diversity of training data, which helps prevent overfitting, and using k-NN for labeling (E) lets you leverage the big set of unlabeled images instead of just sticking to a small labeled set. Pretty sure this is right but open to other ideas.
Question 1 of 20

Premium Access Includes

  • ✓Quiz Simulator
  • ✓Exam Mode
  • ✓Progress Tracking
  • ✓Question Saving
  • ✓Flash Cards
  • ✓Drag & Drops
  • ✓3 Months Access
  • ✓PDF Downloads
Get Premium Access

← Back to all Amazon - AWS exams & practice tests

Scroll to Top