Select AWS case studies with AI / ML Applications
Samsung Electronics needed a better way to predict demand for memory hardware
The company wanted to empower business analysts without coding experience to glean data-driven insights using ML
The traditional methods that the company was using to forecast memory chip demand were sometimes volatile, inaccurate, and didn’t account for new factors, e.g., new applications and devices on the market and environmental factors like the COVID-19 pandemic impacting business
The Memory Marketing team of the Samsung Device Solution division analyzes memory needs for electronics
The team previously based fore-casts on chip demand, customer preferences and regressions; the solution aimed to empower business analysts to create more accurate and less costly demand forecasts
Models trained with Machine Learning (ML) can aid business analysts to better forecast demand and shipments
SageMaker Canvas – No-code fully-managed infrastructure for building, training and deploying AI/ML models for any use case
Increased forecasting accuracy
Generation of Insights in hours, rather than days
Provided Business Analysts with tools to build ML models and generate predictions
Saved time of the data science team
Increased collaboration between the business analysis and data science teams
1. Not exhaustive. Source: AWS, Samsung Electronics Improves Demand Forecasting Using Amazon SageMaker Canvas
Frollo was initially running its databases and core computing platforms on the Heroku cloud platform but wanted more independence in building and deploying ML algorithms
The company was also aiming to optimize its performance–cost ratio for compute-hungry ML instances
Frollo serves Banks and FinTechs providing backend APIs on an SaaS platform: The Frollo Data Enrichment API provides solutions such as categorizing customer transactions and identifying merchants. Together with ML models, the solution can sort consumer spending into more than 60 categories
Frollo required the right ML infras-tructure to provide a high level of data granularity and aimed to optimize its performance-cost ratio through the enablement of ML algorithms
Amazon SageMaker
– Consultancy by AWS Advanced Consulting Partner, Itoc
Ensuring end-to-end API latency of 1,500 milliseconds or less for transaction categorization and merchant identification
Achieving 95% accuracy rate for ML models categorizing transactions
Risk reduction in loan origination
Credit results provided in 20 seconds rather than 3 to 5 minutes
Securing of product reference data for bank
1. Not exhaustive. Source: AWS, Frollo Provides a Fast Track to Open Banking on AWS
The executive team at A2A required a more intuitive way to interact with complex business data
Traditional dashboards lacked the flexibility and easy of use needed for executives to quickly extract specific insights without technical assistance
A2A is an Italy-based energy, water, and waste services provider
A2A wants to put information directly into the hands of directors and executives, so they can act fast and make the best decisions for the business
A2A is exploring services like Amazon Q to speed time to insight with business intelligence dash-boards and natural language interactions
Amazon Q - Generative artificial intelligence (AI) tool
Increase time to insight with business intelligence dashboards and natural language interaction
A2A can deliver smarter services with less impact on the environment
1. Not exhaustive. Source: AWS, A2A Uses AWS and Generative AI to Accelerate Business Knowledge
Mastercard faced increasing challenges in detecting and preventing fraudulent transactions across its vast network
Its conventional fraud detection methods were becoming inadequate, necessitating a more advanced, scalable solution capable of processing and analyzing massive datasets swiftly and accurately
Fraud results in merchants losing billions of Dollars in fraudulent transactions as well as in bad customer experience
It is a challenge for older rule-based systems to keep up
Mastercard uses a number of AWS services with the main focus on three key areas: performance, resilience and security
AWS artificial intelligence (AI) services
AWS machine learning (ML) services
Detect three times the amount of fraudulent transactions
Reduce false positives tenfold
Billions of dollars in merchant savings
Providing a better experience for customers around the world
1. Not exhaustive. Source: AWS, Mastercard Uses AWS AI and ML Services to Detect and Prevent Fraud
MaxAB had been developing its own demand forecasting models in several areas
The company needed to streamline model development and iterate at faster speeds
When the MaxAB data team began working on its demand forecasting models using AWS, it developed them manually in silos
This meant it was difficult to collaborate on model development, resulting in slow model iteration speeds and a lack of model interdependency
MaxAB, founded in 2018 and based in Egypt, offers an app-based and web channel B2B ordering system for grocery store owners
The data team was using AWS for their demand forecasting models, but they needed them to be more accurate
MaxAB benefited from a regional initiative, the AWS Prototyping Program, to help it build a new model development structure
The company has developed its first operational model to track product sales spikes and adjust pricing on the same day
Amazon S3 – Cloud Object Storage
Amazon SageMaker Studio leveraging ML and AI-powered tools
Amazon Athena
Reduced demand forecasting model completion times by over 94 percent, taking it down from 9 hours to just 30 minutes
8x faster model iteration speeds
60% improvement in demand forecasting model accuracy rates
Cost minimized
1. Not exhaustive. Source: AWS, MaxAB Improves Demand Forecasting by 60% for Optimized Pricing Using AWS
To support the innovations related to the ever-evolving online retailing business, YNAP wanted to give its quality assurance (QA) engineers ways to quickly test and launch new features that improve the customer experience
Previously, its 5 QA engineers shared one testing environment, so they had to wait until one tester was finished with a feature before they could begin testing the next one
Also, all of YNAP’s ecommerce sites must be responsive and reliable so shoppers can take advantage of their advanced capabilities
YNAP Group is an online luxury garment retailer
YNAP aimed to include in its online experience, advanced features such as personalized views, virtual fitting, and AI-powered personal shopper suggestion based on individual tastes
YNAP aimed to provide its Quality Assurance Engineers the ability to quickly test and launch new features
A Managed container service can run and scale Kubernetes applications in the cloud to test several features simultaneously
Amazon Elastic Kubernetes service– Managed container service– Scaling of compute resources in seconds compared to hours in previous on-premise systems
Amazon EC2 Spot Instances– Runs fault-tolerant workloads at a discount
Reduced product testing time by 400%
Computing costs cut by 59%
Staff upskilling on cloud-native technologies
Ability to respond faster to customer needs
Number of infrastructure issues requiring immediate response decreased from 2/month to 0
Reduction of infrastructure maintenance man hours from 10 different teams to focus on high-value work
1. Not exhaustive. Source: AWS, YNAP Scales and Innovates Faster to Provide High-End Shopping Experience Using AWS
3Victors processes over one billion travel searched daily and required scalable infrastructure to manage growing data
3Victors needed to easily aggregate the fast-growing volume of data coming in daily from disparate global reservations systems to provide the best business value to travel customers
It grows at a rate of 20 terabytes of data every day, and it would have had to quadruple the size of its infrastructure to support it in an on-premises environment
AWS enables 3Victors to ingest and analyze as much data as they can
Implementing AWS solution to capture and store data for scalability and data analytics
With AWS, 3Victors simplifies big data to help clients increase conversion and return on ad spend while optimizing revenue
3Victors reduces operational costs by moving completely to AWS and running its big data environment on the AWS Cloud
Amazon Route 53 directs vendor content to AWS Elastic Beanstalk
Amazon Redshift for data storage
Amazon Kinesis for data capture, transformation and buffering data
Amazon EC2 Spot Instance to optimize cost
Data access is provided through Amazon API Gate and AWS Lambda
Increased scalability to handle growing data without infrastructure limitations
Real-time processing and faster data provide customers with actionable insights
Help sellers of travel improve content engagement while optimizing revenue
Reduce operating costs: save tens of thousands each month
1. Not exhaustive. Source: AWS, 3Victors used AWS to deliver near-real time travel trend AI insights; 3Victors ingests data from the world’s largest reservations systems to provide data analytics solutions to sellers of travel across the globe
Crypto.com sought to enhance its global reputation by delivering real-time sentiment analysis to its 100 million users
Crypto.com’s developers encountered challenges with the limitations of open-source models and the high cost of self-hosting large language models (LLMs)
They also faced accuracy issues with outputs generated by open-source models, especially when dealing with multilingual news sites
Crypto.com adopted AWS services to achieve scalability and improved data accuracy
By implementing a multi-agent consensus-seeking solution for sentiment analysis on AWS, Crypto.com can efficiently deliver accurate, comprehensive, and localized crypto market insights to its global user base
With Amazon Bedrock, Crypto.com benefits from highly scalable models that automate the generation of insights, saving time and resources
Amazon Bedrock with Anthropic Claude 3 for large language models
Amazon SageMaker to fine-tune customer modles
Amazon Bedrock for ongoing proofs of concept and development
Sentiment analysis response time reduced to 1 second
Increased development efficiency, enabling rapid testing time and deployment
Save time and resources with highly scalable models
Delivers localized insights in 25 languages for a better user experience
Satisfied customer feedback encourages Cryptp.com to continue testing ways to deploy generative AI across the organization
1. Not exhaustive. Source: AWS, Crypto.com delivers accurate sentiment analysis in 1 second with Generative AI on AWS
