Jessie A Ellis
Jun 10, 2026 16:31
Ray Day NYC featured Coinbase, Discord, and Torc Robotics sharing how Ray boosted AI workloads. Highlights embrace Torc’s 90% GPU utilization.
Ray Day NYC introduced collectively engineers from main corporations like Coinbase, Discord, and Torc Robotics to showcase how they’re leveraging the open-source Ray framework and Anyscale to scale AI workloads. Held at One Liberty Plaza, the occasion highlighted real-world manufacturing challenges and the way Ray unlocked new efficiencies, together with Torc Robotics attaining 90% GPU utilization, a dramatic leap from its prior 30-40%.
Torc Robotics: Scaling Multimodal AI for Autonomous Vehicles
Neil Wadhvana, ML Ops Tech Lead at Torc Robotics, outlined the stakes for autonomous trucking—focusing on a $200 billion U.S. long-haul market by 2030 amid a 160,000 driver scarcity. Torc consolidated its fragmented multimodal information processing stack right into a single Python-based engine powered by Ray. This shift allowed independently scaled CPU and GPU node swimming pools, boosting GPU utilization to 90%. Coaching time per epoch dropped from 20 minutes to five, delivering a 4x enchancment with out further {hardware}. Their pipelines now deal with as much as 38 TB of coaching information with ease. For these , Torc is internet hosting a webinar diving deeper into these developments.
Discord: From Open-Supply Ray to Anyscale
Discord’s Senior Software program Engineer Serrana Aguirregaray detailed how the corporate scaled its ML platform from open-source Ray to Anyscale. Initially used to coach advice fashions for 90+ million each day energetic customers, Discord’s adoption of Ray expanded quickly throughout groups dealing with advertisements, security, and content material understanding. Nonetheless, managing a number of Kubernetes-based Ray clusters turned a bottleneck. Transitioning to Anyscale resolved these points with a unified management airplane and declarative configs, enabling extra deal with mannequin growth. Discord’s first deep studying use case improved advert rankings by over 200%.
Coinbase: Scaling Monetary Danger Evaluation
Coinbase’s Aman Choudhary shared how the corporate makes use of Ray and Anyscale to course of hundreds of batch ML jobs each day for monetary danger prediction. After shifting from SageMaker to Ray in 2023, Coinbase reduce prices by 20% and slashed iteration instances from hours to seconds. Subsequent optimizations with Anyscale enabled the group to scale from 3,000 degraded jobs to over 10,000 secure each day jobs, considerably enhancing operational effectivity.
Market Context and Relevance
The improvements on show at Ray Day NYC underscore the rising significance of scalable AI infrastructure as industries combine machine studying into crucial workflows. These developments additionally align with broader developments in blockchain and crypto, the place decentralized platforms like Raydium (RAY) are leveraging developments in distributed computing to push tokenized real-world asset adoption.
Raydium, a Solana-based decentralized alternate, not too long ago surpassed $1 trillion in cumulative buying and selling quantity and reported over $2 billion in tokenized fairness buying and selling quantity. Whereas the RAY token noticed a modest 2.81% value enhance as we speak to $0.5820, the ecosystem faces challenges, together with a $1.34 million exploit of an outdated liquidity pool. The convergence of AI scalability and blockchain know-how may current important alternatives for platforms like Raydium to reinforce their infrastructure.
What’s Subsequent
Ray Day NYC is a part of a broader “Ray on the Highway” initiative by Anyscale, culminating within the Ray Summit in San Francisco on August 24. The occasion will characteristic keynotes, workshops, and product roadmap previews. With Ray’s rising traction throughout industries, the summit is poised to set the stage for the following wave of AI scalability.
Picture supply: Shutterstock
