AWS HPC Blog
Author: Ross Pivovar
Ross has over 15 years of experience in a combination of numerical and statistical method development for both physics simulations and machine learning. Ross is a Senior Solutions Architect at AWS focusing on development of self-learning digital twins, multi-agent simulations, and physics ML surrogate modeling.
Optimizing undersea cables: how Orsted and AWS modeled seabed thermal properties
This post was contributed by Ross Pivovar, Rafał Ołdziejewski, Cindy Xin Qi Lee Offshore wind farms play a critical role in the global transition to renewable energy and clean power generation. But generating electricity is only half the battle—safely and efficiently transporting that power to the grid through undersea cables is equally important. Today, we’ll […]
Leveraging LLMs as an Augmentation to Traditional Hyperparameter Tuning
When seeking to improve machine learning model performance, hyperparameter tuning is often the go-to recommendation. However, this approach faces significant limitations, particularly for complex models requiring extensive training times. In this post, we’ll explore a novel approach that combines gradient norm analysis with Large Language Model (LLM) guidance to intelligently redesign neural network architectures. This […]
Smashing computational barriers: data-driven ball-impact modeling on AWS
Elevate your engineering capabilities with lightning-fast impact prediction. Our new blog post delves into how advanced ML models, like U-Nets and Fourier Neural Operators, are revolutionizing transient response forecasting for critical industries like consumer electronics, automotive, and aerospace. Gain a competitive edge by integrating these cutting-edge techniques.
Optimizing compute-intensive tasks on AWS
Optimizing workloads for performance and cost-effectiveness is crucial for businesses of all sizes – and especially helpful for workloads in the cloud, where there are a lot of levers you can pull to tune how things run. AWS offers a vast array of instance types in Amazon Elastic Compute Cloud (Amazon EC2) – each with […]
Simulating complex systems with LLM-driven agents: leveraging AWS ParallelCluster for scalable AI experiments
How might AI change the rules of the energy game? A new post explores using large language models to power smarter, more adaptive agents in an energy supply chain simulation. Learn how LLMs could enable more nuanced decision-making behaviors.
Harnessing the power of large language models for agent-based model development
Want to build agent-based models without deep expertise? Our latest blog post explores using Claude 3 Sonnet to tap into knowledge and accelerate ABM development.
Using a digital twin for sensitivity analysis to determine sensor placement in a roll-to-roll manufacturing web-line
What’s the best way to select sensors to capture key data for your digital twin without overspending? Check out our latest blog post on using ML and sensitivity studies to optimize sensor selection.
Using a Level 4 Digital Twin for scenario analysis and risk assessment of manufacturing production on AWS
This post was contributed by Orang Vahid (Dir of Engineering Services) and Kayla Rossi (Application Engineer) at Maplesoft, and Ross Pivovar (Solution Architect) and Adam Rasheed (Snr Manager) from Autonomous Computing at AWS One of the most common objectives for our Digital Twin (DT) customers is to use DTs for scenario analysis to assess risk […]
Using Fleet Training to Improve Level 3 Digital Twin Virtual Sensors with Ansys on AWS
AWS is developing new tools that enable easier and faster deployment of level 3/4 digital twins. This post discusses how a fleet calibrated level 3 digital twin can be cost effectively deployed on AWS Cloud.
Deploying Level 4 Digital Twin Self-Calibrating Virtual Sensors on AWS
Digital twins can be hard if they deviate from real-world behavior as real systems degrade and change over time. Today we’ll show digital twins that calibrate on operational data, using TwinFlow on AWS.