Scarlet Platform
Cut the learning curve and integration effort
Run smart and fast at reduced costs
AI Training, Inference and Analytics in the Cloud
TensorFlow and PyTorch integration
Scarlet AI
Inference & Analytics
Scarlet AI Training,
Inference & Analytics
No data sharing. Hourly and usage-based pricing
Craft your AI and data strategy
Train and visualize TensorFlow and PyTorch AI models with ease and flexibility. Simply upload your training scripts to the platform and create training jobs that are automatically managed. With one click, you can export your trained models for inference.
It supports scheduling, multiple run history, instance type choice, per run visualization and cost/saving estimates.
Easily run AI models for image, text and time series data on your S3 and GCS buckets. You can choose from the models you trained on the platform, upload your own pre-trained TensorFlow or PyTorch models or use one of the pre-integrated cloud models (AWS Rekognition, GC Cloud Vision, GC Cloud Natural Language and SageMaker integration). You can search and visualize the AI analytics results, and use them for custom notifications and auto-labeling.
Explore your cloud data through data source and global level dashboards. Get meaningful insights at a glance from relevant metrics and historical states, and visualize the distribution of your data by content type, age, size and more. Learn how to optimize your space utilization and how your data evolves over time.
Easily build complex queries based on object metadata, AI predictions and cross-object analysis. Retrieve 1000s of results in milliseconds from your entire dataset or just a subset. Dive deeper into the search results by accessing detailed information for each object and export reports for further processing.
Get notified when your data meets criteria you define, for example when certain types of images are detected in some of your data sources.
Customize your settings based on analytics results and the scope and the severity of the alert.
Apply targeted actions to selected data using flexible policies that you can run on-demand or schedule to run automatically. You can for example organize your data based on the AI analysis results, remove duplicate objects, copy or move certain data to a new location or rename documents that meet certain criteria.
Testimonials
DataMacaw’s Scarlet platform has been a joy to work with, and has transformed how we train models at Patronus AI. The best part is probably being able to train models on cost-efficient AWS spot instances, while retaining saved model weights and results in a centralized GUI. It’s easy to manage training jobs, and is more cost-effective than AWS’s SageMaker machine learning platform.
We also love its native TensorBoard integration, handy job completion notifications, and helpful support. I would recommend DataMacaw to any team with a complex machine learning workflow.
It is a very impressive product, Scarlet did analyze my S3 buckets and present very coherently. The UI is very polished and easy to get around.
Unlock the value of your unstructured data
We combine powerful integration and intelligent resource management to give you high performance at low running cost
Use cases
-
Organize your unstructured data
Classify and label your unstructured data from your cloud buckets, such as pictures using a vision AI model, and organize them by category in a target location of your choice
-
Detect sensitive content
Extract text content from different types of documents, label them using an NLP model and receive notifications if certain type of content is detected
-
Triage large amounts of data fast
Quickly ingest, analyze using AI and triage large amounts of regularly generated data, such as log files. Retain only the relevant objects in a target location of your choice
-
Clean up duplicate data automatically
Detect duplicate objects across multiple buckets and clean up unneeded copies using a rule of your choice, for example keeping only the most recent copy or the copy in a specific bucket.
Scarlet Features
AI training
Scarlet offers training of TensorFlow and PyTorch models with ease and flexibility on dynamically managed CPUs and GPUs. It supports scheduling, multiple run history, instance type choice, per run hyper parameter tuning, per run visualization and cost/savings estimates.
AI inference and analytics
Scarlet allows running AI models trained on the platform, importing TensorFlow and PyTorch models, or using pre-integrated cloud models. It also prepares your data e.g. extracts text from different types of documents. With its intelligent metadata collection and efficient indexing, you get deep insights into your data.
Instant search and notifications
You can perform instant search across your entire data or just a subset and get full reports of your search results. You can also add your custom notifications to detect if and when your data meet a certain criteria. This can be particularly useful when combined with AI analysis and automated labeling, for example to spot the presence of sensitive content in your data.
Flexible policies
You can customize your data management strategy to your specific needs by adding policies that enable you to apply tactical actions to targeted data. You can run policies on-demand or schedule them to start automatically, for example to organize your data by category based on the AI analysis or to remove your duplicate data.
Scalable and cost effective
The platform leverages cloud storage and cloud processing capabilities, and the full potential of its deployment environment. Its patent pending architecture combines powerful distributed processing with highly responsive resource management to offer maximum performance at a fraction of the costs incurred otherwise.
Unified analytics and operation
The platform offers seamless integration of AI training/inference and analytics, seamless operation across sources from different cloud domains, and seamless policies across different regions. You can analyze and manage your Amazon S3 and GC Storage from a single console. The unified visibility and control enable smart efficient management of your data.