Suggestions
Raghvender Arni
Raghvender Arni is a highly accomplished technology professional with extensive experience in cloud computing, artificial intelligence, and software architecture.12 He currently leads the Cloud and AI Innovation team (CAIT) at AWS, focusing on accelerating and optimizing business outcomes through cloud and AI adoption across industries.1
Prior to his role at AWS, Arni held the position of Head of Platform Architecture at Pivotal Software, Inc., where he led a global team of technologists driving digital transformation for major Global 2000 companies.1 His career journey includes:
- Education: Master of Science from Rutgers University, Bachelor of Science in Electrical Engineering from Birla College of Engineering, and further studies at Villanova University.
- Early career: Began at Bell Labs Research after graduate studies in Computer Engineering.12
- Leadership roles: Held various senior positions at Pivotal Software, Inc., including Senior Director of Platform Strategy and Advisory Field Architect.
- Other experiences: Worked as Principal Solutions Architect at Progress, Senior Solutions Architect at Cast, and Solutions Architect at Danucom.
- Entrepreneurial ventures: Co-founded Dset and served as a Product Manager at Danucom.
Arni's expertise spans cloud architecture, AI/ML technologies, IoT, robotics, and advanced data computing.2 His LinkedIn username is rarni.3
Highlights
𝗧𝗟;𝗗𝗥: As GenAI hype stabilizes, enterprises are viewing AI (predictive and generative) more holistically. They're also realizing that AI adoption strategies are very similar to cloud (and similar transformational tech) adoption strategies. To succeed with AI in the enterprise, start by reviewing your cloud playbook!
𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 is just one branch of broader AI. There are so many use cases and techniques where 𝗣𝗿𝗲𝗱𝗶𝗰𝘁𝗶𝘃𝗲 𝗔𝗜 (for the lack of a better word) is the only solution or the better and cheaper solution – optimization, forecasting, anomaly detection etc with regression models, random forests, gradient boosting etc.
𝗔𝗰𝘁𝗶𝗼𝗻 𝟭: 𝗟𝗼𝗼𝗸 𝗮𝘁 𝗔𝗜 𝗵𝗼𝗹𝗶𝘀𝘁𝗶𝗰𝗮𝗹𝗹𝘆, 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻 𝗶𝗳 𝗚𝗲𝗻𝗔𝗜 𝗶𝘀 𝘁𝗵𝗲 𝗵𝗮𝗺𝗺𝗲𝗿 𝗳𝗼𝗿 𝗲𝘃𝗲𝗿𝘆 𝗻𝗮𝗶𝗹.
Cloud adoption continues to grow (https://gtnr.it/3WZyULV) in the enterprise and enterprises have built some great artifacts for transformation: 1/Cloud Strategy and Roadmap 2/Cloud Maturity Model 3/Cloud Centers of Excellence 4/Cloud Architecture Patterns 5/Cloud Business Value/Model generation 6/ Cloud Governance & Security 7/Cloud Skills and Transformation
AI is going through the same journey especially as 𝗔𝗜 𝗵𝗮𝘀 𝗮𝗰𝗰𝗲𝗹𝗲𝗿𝗮𝘁𝗲𝗱 𝗱𝘂𝗲 𝘁𝗼 𝘁𝗵𝗲 𝗽𝗼𝘄𝗲𝗿 𝗼𝗳 𝘁𝗵𝗲 𝗰𝗹𝗼𝘂𝗱. Matt Wood spoke about the S-curve of innovation (https://lnkd.in/eWzEfHJD), with Cloud we are more than half-way through, with AI we are just getting started!
Take a lot of lessons from your cloud success as you kick off your AI journey. We at Amazon Web Services (AWS) have a 𝘁𝗲𝗮𝗺 𝗼𝗳 𝗖𝗹𝗼𝘂𝗱 𝗮𝗻𝗱 𝗔𝗜 𝗮𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘀 who can help you with your 1/AI Strategy and Roadmap 2/AI Maturity Model 3/AI Centers of Excellence 4/AI Architecture Patterns 5/AI Business Value/Model generation 6/ AI Governance & Security 7/AI Skills and Transformation – see what I did there :). Yes, AI has a lot of its own uniqueness but there is a ton of similarity with cloud adoption.
𝗔𝗰𝘁𝗶𝗼𝗻 𝟮: 𝗕𝘂𝗶𝗹𝗱 𝗮𝗻 𝗔𝗜 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝘆 𝗮𝗻𝗱 𝗿𝗼𝗮𝗱𝗺𝗮𝗽 𝘁𝗵𝗮𝘁 𝗯𝘂𝗶𝗹𝗱𝘀 𝗼𝗻 𝘆𝗼𝘂𝗿 𝗖𝗹𝗼𝘂𝗱 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝘆 𝗮𝗻𝗱 𝗿𝗼𝗮𝗱𝗺𝗮𝗽
TL;DR: As more enterprises go to production with AI its interesting to see how their 𝗺𝘂𝗹𝘁𝗶-𝗺𝗼𝗱𝗲𝗹 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝘆 is evolving. It’s usually a combination of 𝗺𝘂𝗹𝘁𝗶𝗽𝗹𝗲 𝗳𝗿𝗼𝗻𝘁𝗶𝗲𝗿 𝗺𝗼𝗱𝗲𝗹𝘀 combined with some 𝗰𝘂𝘀𝘁𝗼𝗺 𝗳𝗶𝗻𝗲-𝘁𝘂𝗻𝗲𝗱 𝘁𝗮𝘀𝗸-𝗳𝗼𝗰𝘂𝘀𝗲𝗱 𝗺𝗼𝗱𝗲𝗹𝘀 with a 𝗺𝗼𝗱𝗲𝗹 𝗿𝗼𝘂𝘁𝗲𝗿 in front that routes between the models balancing 𝗮𝗰𝗰𝘂𝗿𝗮𝗰𝘆, 𝗰𝗼𝘀𝘁 𝗮𝗻𝗱 𝗹𝗮𝘁𝗲𝗻𝗰𝘆. Cisco Webex and Zoom show the way and Amazon Web Services (AWS) SageMaker is the star in the Cisco story!
We have too many meetings in our corporate lives but companies like Cisco Webex and Zoom are trying to reduce the pain a bit by providing 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱 𝗺𝗲𝗲𝘁𝗶𝗻𝗴 𝘀𝘂𝗺𝗺𝗮𝗿𝗶𝗲𝘀, 𝗮𝗰𝘁𝗶𝗼𝗻 𝗶𝘁𝗲𝗺𝘀 𝗮𝗻𝗱 𝗶𝗻𝘀𝗶𝗴𝗵𝘁𝘀 all using the power of AI.
Interestingly both of them have a similar strategy of routed & federated multi-model AI. Let’s look at Cisco Webex solution:(https://lnkd.in/en3hyAgs)
• Deployed 𝗳𝗶𝗻𝗲-𝘁𝘂𝗻𝗲𝗱 𝗟𝗹𝗮𝗺𝗮 𝗮𝗻𝗱 𝗕𝗘𝗥𝗧 𝗺𝗼𝗱𝗲𝗹𝘀 for 𝘀𝗽𝗲𝗰𝗶𝗳𝗶𝗰 𝘁𝗮𝘀𝗸𝘀 like topic clustering, topic labeling, intent extraction etc. All these were built and deployed on SageMaker (more below) • They used 𝗳𝗿𝗼𝗻𝘁𝗶𝗲𝗿 𝗟𝗟𝗠𝘀 on Amazon Bedrock for call 𝘀𝘂𝗺𝗺𝗮𝗿𝗶𝘇𝗮𝘁𝗶𝗼𝗻 • An 𝗟𝗟𝗠 𝗣𝗿𝗼𝘅𝘆 𝗿𝗼𝘂𝘁𝗲𝗱 𝗿𝗲𝗾𝘂𝗲𝘀𝘁𝘀 between various models across Bedrock, SageMaker and other providers
Amazon SageMaker team and Cisco Webex together did some amazing things:
• They 𝗿𝗲𝗱𝘂𝗰𝗲𝗱 𝗱𝗲𝗽𝗹𝗼𝘆𝗺𝗲𝗻𝘁 𝗰𝗼𝘀𝘁𝘀 by 50% and latency by 20% • They 𝗿𝗲𝗱𝘂𝗰𝗲 𝗮𝘂𝘁𝗼 𝘀𝗰𝗮𝗹𝗶𝗻𝗴 𝘁𝗶𝗺𝗲𝘀 by up to 40% and auto scaling detection by 6 times • They simplified deployment of c𝗹𝗮𝘀𝘀𝗶𝗰𝗮𝗹 𝗠𝗟 𝗮𝗻𝗱 𝗟𝗟𝗠𝘀 together! • For Kubernetes deployments use of Kubernetes operators to 𝗿𝗲𝗱𝘂𝗰𝗲 𝗟𝗟𝗠 𝗱𝗲𝗽𝗹𝗼𝘆𝗺𝗲𝗻𝘁 𝗰𝗼𝘀𝘁𝘀 𝗯𝘆 𝟱𝟬%
Kudos to the SageMaker (Ankur Mehrotra) & Cisco Webex teams.
Also read how Zoom approached their Federated AI solution from their CTO Xuedong Huang: https://lnkd.in/eDXSid3x
𝗔𝗰𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗖𝗧𝗢𝘀, 𝗖𝗔𝗜𝗢𝘀: Invest in 1/Model Fine-tuning for tasks/domains and 2/Model Routing
