BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//DevOpsChat//Events//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
UID:01a1989d-2ca2-72f1-9485-a58068de521d@www.devopschat.co
DTSTAMP:20261106T131836Z
DTSTART:20200124T203000Z
DTEND:20200124T213000Z
SUMMARY:AI on Cloud
DESCRIPTION:This is a FREE meetup. Participants from different experience l
 evels are invited.\n\nAgenda:\n\n10:00 AM - 10:30 AM - Registration\n\n10:
 30 AM - 11:15 AM - "Building and Deploying AI/ML models on Cloud using Clo
 ud-Native way" by Vishwanath\, GE Oil & Gas\n\nAbout Vishwanath: https://w
 ww.linkedin.com/in/vishwanath-shankar-8233418/\n\nAbstract: The biggest ch
 allenge enterprises face these days is to scale and run the AI/ML models d
 eveloped in house. Also\, a hybrid cloud setup makes the challenge even bi
 gger. In this talk\, we will be going through the basics of an ML workflow
  by going through a practical data set\, build out an ML model\, deploy an
 d scale the model using cloud technologies.\n\n11:15 AM - 11:50 AM - “Th
 anos: Achieve highly scalable and global monitoring with Prometheus” by 
 Kirti Ranjan Parida\, Lead DevOps Engineer at Informatica\n\nAbout Kirti R
 anjan: https://www.linkedin.com/in/kirti-ranjan-parida/\n\nAbstract: In In
 formatica\, cloudtrust supports multiple infrastructures for product teams
  spanning across 3 major cloud providers (AWS\, Azure\, GCP) and multiple 
 regions. Hence we need highly scalable and resilient monitoring system whi
 ch not only provide system uptime but also historical data about systems a
 nd application. We also need a single pane of the window to know what is h
 appening with our systems. To solve this\, we are using a combination of P
 rometheus/Thanos/Grafana for monitoring our systems(~2000 Cloud instances\
 ,~20 Kubernetes clusters ~3k containers and expect the Kubernetes infrastr
 ucture to grow by 5x at the end of 2020 ). Thanos is an open-source projec
 t by Improbable to seamlessly transform existing Prometheus deployments in
  clusters around the world into a unified monitoring system with unbounded
  historical data storage.\n\n11:50 AM - 12:00 PM - Break & Networking\n\n1
 2:00 PM - 12:45 PM - "Serve ML Models (low-latency prediction systems) at 
 Scale in the Cloud and at Edge" by Srinivasa Rao\, Cisco\n\nAbout Srinivas
 a: https://www.linkedin.com/in/aravilli/\n\nAbstract: Low Latency predicti
 on serving systems are the key to serve and scale ML/AI applications at th
 e edge and in the cloud. In this talk\, I will present various challenges 
 involved in scaling ML applications and serving prediction systems and pre
 sent open-source frameworks like Clipper\, Kubeflow\, ml-flow which can be
  used in order to overcome the challenges. I will walk through a sample cl
 assification use case ( Phishing detection ) to deploy\, scale and serve u
 sing various frameworks and explain challenges in detail.\n\n12:45 PM - 01
 :00 PM - QA & Networking\n\nSponsored by - Microsoft\nVenue partner - Info
 rmatica\n-----------------------------------------------------------------
 -----------------------------------\nTerms & Conditions: You understand an
 d accept that you will abide by the KonfHub's Code Of Conduct (https://kon
 fhub.com/codeofconduct.html).\n\nhttps://www.devopschat.co/meetings/ai-on-
 cloud
URL:https://www.devopschat.co/meetings/ai-on-cloud
END:VEVENT
END:VCALENDAR