Alluxio

Is your data infra struggling to meet the performance demands of #AI and #machinelearning workloads? This technical whitepaper demonstrates critical architectural considerations for optimizing data access in enterprise AI infrastructure. You will learn: 🐢 Common data access challenges that slow down your GPUs 🔍 Why NFS/NAS may not be your best choice 💯 Reference architecture on AWS and benchmarks test results Read Now: https://buff.ly/49aLMDD #gpu #GTC #GTC24 #nfs #nas #modeltraining #llm #storage #compute #machinelearning #AI #ML #genAI

📹 In this video, Hope Wang, developer advocate at Alluxio, unravels the evolution and rising popularity of open-source data lakes for big data analytics and AI workloads. Gain insights into: 💡 The evolution of data platform architectures 📈 Trends in cloud-native approaches ✨ The importance of efficient data retrieval and bandwidth for analysis Click to watch now: https://buff.ly/3QkR9sz #DataCouncil #data #io #ai #modeltraining #machinelearning #storage #compute #anlytic #cloud

Prefill in LLM inference is known to be resource-intensive, especially for long LLM inputs. While better scheduling can mitigate prefill’s impact, it would be fundamentally better to avoid (most of) prefill. Mark your calendars for “Reducing Prefill for LLM Serving in RAG”, an enlightening session by Junchen Jiang, Assistant Professor of computer science at University of Chicago where He will introduce how to speed up prefill delay while maintaining the same generation quality by improving the loading process of the reused KV cache. Learn More: #data #ai #llm #gpu #NVIDIA #modeltraining #machinelearning #deeplearning #storage #compute #caching #dataplatform #cloud

Don’t miss our coming up AI/ML Infra Meetup, a hybrid community event happening at Uber Sunnyvale and on zoom. Introducing the last talk for the night, “My Perspective on Deep Learning Framework”, presents by Xiande Cao, Senior Deep Learning Software Engineer Manager at #NVIDIA. From Caffe to MXNet, to PyTorch, he will share his perspective on the evolution of deep learning frameworks. Learn More:

Slow #data access and low #GPU utilization can bottleneck end-to-end machine learning pipelines as training data volume grows and when large model files are more commonly used for serving. These challenges are prevalent when using popular frameworks like #PyTorch, #Ray, or #HuggingFace, paired with cloud object storage solutions like #S3 or #GCS. Join us at AI/ML Infra Meetup at Uber where Lu Qiu and Siyuan Sheng will offer comprehensive insights into improving speed and GPU utilization for model training and serving. RSVP:

👏 We are honored to be recognized as one of the leading companies in the sphere of big data management and integration tools, as acknowledged in part 4 of the 2024 CRN Big Data 100 list. Resides between compute and storage, Alluxio brings data closer to accelerate heavy duty analytics/ai workload. Read this article to learn more:

🙌 Can't wait to meet you all at Trino Fest 2024!!

🚗 #Uber builds and maintains one of the largest scale ML infrastructure, with over 1000 pipelines daily, for training an extensive number of models being used across various aspects of the business. Join Qiushen (Eric) Wang, Software Engineer at Uber’s Michelangelo team at our AI/ML Infra Meetup for his session “Optimizing Data Pipeline on Uber’s ML Platform” to learn challenges introduced by such a massive scale of data pipeline. You can expect: 🚀 Enhancing Data Access Speed & Efficiency 🔝 How to maximize CPU & GPU utilization in training epochs 🤝 Shared data infrastructure for collaborative project efficiency #data #ai #llm #gpu #NVIDIA #modeltraining #machinelearning #deeplearning #storage #compute #caching #dataplatform #cloud

The third part of our multi-cloud series webinar, “Cloud-Native Model Training on Distributed Data” is only one week away! Don’t miss the chance to embrace hybrid or multi-cloud architecture for large-scale #analytics and #AI workloads. Save your spot now: https://buff.ly/3U1rq9F #cloud #multicloud #hybridcloud #compute #storage #modeltraining #machinelearing

🌟 Our exciting hybrid community event, AI Infra Meetup will be back on Thursday, May 9! 📢 Co-hosted by Alluxio and Uber, we bring leading AI/ML infrastructure experts to Uber Sunnyvale, where they will give talks and share insights about optimizing data pipelines, accelerating model training and serving, and designing scalable architectures. Mark your calendar for this premier opportunity to engage and discuss the latest AI/ML trends with industry professionals from Uber, Nvidia, UChicago, and more, and immerse yourself with learning, networking and conversations. 🍺 Food and Drinks are on us! 🍕 Learn More: https://buff.ly/4b2XQI1 #meetup #al #ml #machinelearning #infra #storage #compute #gpu #llm #modeltraining