Save, migrate and resume
for compute jobs.
Never lose your work again.
Cedana automatically moves jobs across instances and vendors, without interrupting progress, breaking anything or requiring code modification.
Democratizing access to compute
We want to democratize cloud compute across the world. We want to free builders from the the increasing barriers between them and compute, while increasing resource efficiency. Developers should be able to set their preferences for cost, performance and other criteria and let Cedana take care of the rest.
Similar systems to Cedana are in use at OpenAI, Microsoft Research, Meta and others. We help bring them to everyone else.
Access to compute is increasingly a bottleneck to research and progress. There is a growing divide across academia, nations, companies and individuals that separates those with access and the ability to scale compute resources and those that don't. The importance of efficient compute utilization grows with the difficulty of modeling real-world complexity.
We solve these issues by moving your compute in real-time across the world, building a global compute exchange.
Enhance Efficiency
Assign reliability, performance requirements and SLAs from individual jobs to entire tasks.
Target price/performance criteria based on the workload (e.g. inference, training, rendering, simulation).
Reduce compute costs 20%-70%. Improve utilization by migrating between VMs in realtime to optimize price & performance.
Vendor-agnostic compute arbitrage. Compute jobs can be deployed across vendors. Bring your clouds or use ours.
Failure resilience ensures your work runs for as long as it needs to, with no wasted or idle hours.
We support pause/migrate/resume across GPUs with more features to come.
Team
Our team has built real-world robotics systems that interact with humans at 6 River Systems/Shopify. This includes highly available cloud systems that control and manage robot fleets, behavior trees for advanced navigation within warehouses and OTAs in difficult, network constrained environments.
We’ve built large-scale automated compute vision systems at MIT including supporting compute infra to manage large pipelines. We’ve led the development and commercialization of NLP for clinical workflows used in the delivery of patient care. Our team’s publications span computer vision, computer graphics, robotics optimization, and spacecraft/aerospace controls, with patents in AI use cases for grid energy management, optimal battery control and healthcare.
© 2023 Cedana Systems Inc.