Omar Khan

Here’s how Amazon’s homegrown chips are helping customers power complex AI and general compute workloads to process enormous amounts of data at a lower cost.

3

With NVIDIA AI Foundry, enterprises and nations can supercharge #generativeAI with a service spanning curation, synthetic data generation, fine-tuning, retrieval, guardrails, and evaluation to deploy custom AI at Meta Llama 3.1 NVIDIA NIM Microservices.

52

1 Comment

Discover how Oracle Cloud's computer vision and data science services improve the speed of #AI predictions by utilizing the NVIDIA Triton Inference Server: https://lnkd.in/dDwTSCQy

2

What an amazing few days at #SnowflakeSummit2024. A huge thank you to all our UK customers and partners that travelled to San Francisco for the week with Snowflake. Also a massive thank you to our amazing Marketing team for putting on yet another tremendous event. It was amazing to see the new announcements and how Snowflake's architecture is revolutionising AI applications; Iceberg tables are now in GA, Polaris Catalog is now available to support open storage architectures and we are adding an Internal Marketplace along with more AI governance to Snowflake Horizon. These were just a few of the many announcements. With our best-in-class enterprise data foundation and its ability to unlock interoperability with open data and internal collaboration we have exciting times ahead as we unlock the power of the "AI Data Cloud". Finally, a huge thank you to all our Snowflakes who consistently drive customer awareness and the art of the possible. A consistent theme of the week was - "Don't let your imagination limit the endless possibilities that Snowflake's AI Data Cloud can deliver to your organisation". Adam Metherell Florence Storr Andy Westcott Michael Foster Benjamin Tucker Rebecca Sopp Jonathan Nicholson Tom Doyle Rob James Josh Foster-Timms Alex Lambert Caroline Otterheim Lydia Carter Louisa Kierath Alistair Jayne-Coupe Corey Cummins Maxime Fremy James Petter

77

I am not sure if I am scared or excited about these models by OpenAI. Eitherway, I guess i have figured out where a programmer fits in this new era where AI can code so well! I tried these models at length today with Mandeep Singh. They are amazing. Within span of 3-4 hours we were able to execute two project ideas and three of really complex coding tasks. All of this would have taken a team of 2 developers many days to complete. It was scary, it was exciting and very insightful. I literally couldn't hold my excitement and ran out of my limit on the preview model. While it lasted, it was great but It would very often produce wrong code. 90% of it would be correct, but then there would be something that it misunderstood or messed up which would have the whole project fail. As programmers we were able to catch the error and correct it within seconds. I don't think a non dev would have been able to do that (yet). For example, we asked it to create a custom playlist creator based on a BaniDB API by Khalis Foundation, we gave all the function and engineering requirements in detail. We were trying to impress it with our engineering management chops! We copy pasted the code and it didn't work. It misunderstood how the API worked. Or may be I didn't prompt it correctly. It also misunderstood one of the requirements. Maybe it was too abstract for the model. But we didn't give up, In the second prompt we mentored it, like we would mentor a new engineer. We gave it some instructions and voila! it gave the correct output in second try. Within a span of 20 minutes , we were able to create a small web app that would otherwise have taken an engineer hours to do. It literally 10xed our productivity. However i am pretty sure if you give the same tool to a non-programmer the results would still be exemplary but not the same. Maybe I am just trying to grasp to whatever hope I have left :D but for now I believe it's more of a tool to increase dev productivity rather than a tool to replace devs altogether. What are your thoughts?

26

3 Comments

Thats the exact explanation I gave to a group of investors last week. (… it’s just Precision/Recall, attributes vectorization as “embedding”, K-Nearest Neighbor, User Search Intent, DAG, ML algorithms, etc all over again …). LOL and they asked what’s new this time around. And I was like not much, ML is still the backbone, and AI is the inference and actions based on the model. It works better when there is a feedback loop in a closed system. But the current state of the art like ChatGPT is based on scraping the Internet (garbage in, garbage out). Kinda like early search engines. What is a better analogy is that NVIDIA/Snowflake etc are selling tools, aka pickaxes to prospectors. The gold is up to you to find. And the opportunity right now is to monetize industry specific (healthcare, travel, financial services) clean data to build models with (aka a treasure map for the prospectors). From an enterprise perspective, what’s out there today aka generative AI isn’t very useful. Too much hype. To build useful engines will require an effort similar to search revolution 2 decades ago. From

16

3 Comments

When I speak to customers on AI, they often ask where do you even start? This was a great conversation of matching a project to the AI resources. We also dive a bit into what it takes to build an AI Cloud

16

Scaling RAG applications to large datasets means scaling your knowledge retrieval system. This pushes on many aspects, from the ability to handle large data volumes, to performance, economics, and result quality. In this show of Microsoft Mechanics, Jeremy and I explore this topic and go over how we address these challenges in Azure AI Search. We go deep into scaling vector search using vector compression techniques such as quantization and MRL, and how rescoring brings quality back up. We also talk about how at scale relevance is more important than ever, and how our Hybrid Search and Semantic Ranking capabilities help create generative AI applications with better quality responses. https://lnkd.in/gBXtSNqM

372

4 Comments

Infinite Reality announced our strategic partnership with Google! This 5-year partnership will help accelerate iR's offerings through Google Cloud's world-class cloud infrastructure and AI offerings. This partnership has far-reaching benefits for all of Infinite Reality's business operations and it's significance is exemplified through iR Studio. iR Studio is a revolutionary product where immersive content is authored and published for direct delivery through a web browser. It is the single place where the most amazing technologies converge to provide experiences unachievable with the web technologies of the past. Powered by Google Cloud, we are building out product offerings to transform the way people interact with such as how SMBs approach #eCommerce through immersive shopping using #iREngine and generative #AI agents to increase the efficiency and personalization of their customers' shopping experiences. This is just the beginning and the possibilities for our general purpose platform extend beyond shopping into #DigitalTwins, #CustomerService, #Education through our unique ability to recreate and augment digital media in a way reserved only for the most well-funded businesses in the past. Our #Sports and #Entertainment use cases allow every event to turn into a global experience with limitless attendance, audience engagement, and exclusive content. Stay tuned and get ready to see how we harness the power of these technologies to empower our customers to build experiences they never thought possible. Infinite Reality's leadership team has been working tirelessly to make this deal happen to enable this future. Thank you in particular to Rachel Jacobson (Rosen), Melanie Wallner, our new partners at Google Cloud, and our CEO, John A. for the leadership to make this happen. Full release: https://lnkd.in/gUyeEYcP

54

4 Comments

At MongoDB, we see the magic of AI through two lenses: one is incredible user experiences and the second is the intricate, brilliant work of developers who bring these experiences to life. To the world, AI apps are modern-day genies, granting wishes with a snap of a photo or a simple voice note. But developers see the complexities of working with various data types including text, image, and audio. This is why we channel our efforts to empower developers with seamless data unification capabilities, flexible integration options, robust security features, and an intuitive, enjoyable development environment. The potential applications of AI are vast and largely unexplored. If we offer our developers, the right tools and platform, there are so many innovations we are yet to be surprised by. #MongoDB #developer #artificialintelligence

27

Business leaders are looking to Microsoft to understand how #Copilot and agents will transform work—and we want to help. As an engineering leader who supports our sales organization, I’m looking to drive more revenue, so we’re leveraging AI and Copilot across our teams to help increase the pipeline and deal velocity, improve conversion rates, and ultimately improve the seller experience. But that’s just the beginning. We see Agents as being game changers for the sales function. We will have autonomous agents prospect and qualify leads at scale on behalf of sellers and help make them more productive and effective. Discover on #WorkLab how AI can future-proof your organization and keep you ahead of the curve. #AI #Sales #Innovation #Microsoft #FutureOfWork https://lnkd.in/gmJUbn2N

41

From revealing our new LEED silver certified Seattle headquarters, to turning AI hype into real business impact, 2023 kept us on our toes. Explore how we stayed grounded in our annual report: https://slalom.ws/4d4pQvg

1

Microsoft's AI model, Aurora, quickly predicts levels of six major air pollutants worldwide, including carbon monoxide and particulate matter. Read more in this article from Nature. #msftadvocate #ai https://lnkd.in/g-AppzdB

3

Listening to the Microsoft Investor call and like Alphabet Inc., lots of AI-powered growth, particularly in Azure, which was a major driver of growth this quarter (and 20% YoY). Here are a couple of my observations. 1) Many product lines delivered solid YoY growth and Copilot was positioned as a big part of that product line growth. This growth was both incremental, such as the upsell of Office 365 with Copilot, and integrated, such as offerings like Search or Server. It will be interesting to see how revenue and profits will be attributed to CopIilot and AI in the future since it has been used so pervasively. Needless to say, it's a big part of the current and future story. 2) Azure growth was very solid, and while I maintain a concern that unit prices are driving growth vs unit sales (which I said with GCP as well). However, both $10M and $100M contracts were up. So, even if growth plateaus as the cost of AI goes down, at least a good deal of revenue is consolidated in big customers. 3) AI usage was another highlight. GE has deployed a self-service application to 52K users. There was also a 55% increase in usage of GitHub Copilot. In addition to these highlights the number of times leadership mentioned "fastest growth ever" was notable. 4) It was also interesting how much time Satya Nadella spent talking about the GitHub announcements this week.. I like how they are opening up GitHub to a broader swath of users with its new Spark offering. It may be too early to say that AI has "proven it." This is certainly solid progress.

7

Open source has shaped the tech industry for enterprise cloud computing, the innovation with the thousands of our customers - most of the F500 - and our partner ecosystem for the last 30 years. And our AI-driven portfolio for develop, deploy, manage, and automate is actually open. Google InstructLab and the Granite model and check out RHEL AI and OpenShift AI. redhat.com. #redhat Meta CEO Mark Zuckerberg recently shared his vision for the future of technology during a live recording of the "Acquired" podcast in San Francisco. He emphasized the importance of open-source models in the ongoing competition within the tech industry. According to Zuckerberg, the debate between open and closed models is more than just a business strategy; it’s an ideological battle that could shape the future of how technology is developed and used. https://lnkd.in/eWHFJdct

4

Jay is spot on here "One of the biggest barriers to AI innovation is access to clean and trustworthy data. The truth is, while you can have a great AI/ML model, without contextual, real-time data from your business, the model is useless for most applications. Connecting AI models to enterprise data in real-time is one of the more challenging problems to solve." Many may think they are "doing AI" but until this bigger, more challenging problem is solved, AI will have little impact in the enterprise.

1

Graphs offer unparalleled potential for uncovering and leveraging intricate data relationships, driving deeper insights. However, the management of separate graph databases has historically hindered the realization of comprehensive, integrated analytics. Spanner Graph promises to revolutionize this landscape by seamlessly incorporating graph capabilities, thereby unlocking new avenues for data-driven value creation. #graphdatabases #googlecloud https://lnkd.in/dbkYMGYA

24

1 Comment

Join Google Cloud CEO Thomas Kurian and industry leaders to discover how AI is reshaping businesses across the globe. Learn from leading customers how they're harnessing Google's generative AI to drive innovation and gain direct insights into the latest AI advancements and their practical applications. Register now!

6

Oracle today announced the general availability of HeatWave #GenAI, including in-database LLMs, an automated in-database vector store, and the ability to have human-like contextual conversations in natural language. https://lnkd.in/gqZ2ebDS

1