Ari Backholm

Learn how PerfectScale + Karpenter can provide an additional 30 to 50% in cost reductions on top of what can be achieved with Karpenter alone. https://hubs.la/Q02ywfZ90 #kubernetes

7

💡 Why do we need so many evaluation tools for LLMs? As engineers, we build 'production-ready' LLM products using these metrics. But what happens next? How do we maintain control and ensure reliability? At Feedback Intelligence, we’ve crafted a cookbook to keep your LLMs reliable and aligned with user expectations. 🍲 📖 Give it a read and let’s chat!

16

Falkonry (Suhas Mehta and Ashis Khan) and iba AG (Eric Snyder) will present how steel makers are pivoting to AI & Data driven decision support system at AISTech 2024 today (May 6, 2024) - please join us if you are interested. I co-authored this paper with Eric Snyder of iba AG.

66

5 Comments

Power that powers LLMs ! After a decade of flat power growth in the U.S, the rapid growth of AI and related industries is projected to significantly increase U.S. electricity demand by up to 20% by 2030, according to a Wells Fargo analysis. AI data centers alone may add 323 terawatt hours of electricity demand, seven times greater than New York City's current consumption of 48 terawatt. Data centers will account for 8% of total U.S. electricity consumption by the end of the decade. Interesting article: https://lnkd.in/gPjFiSt5

100

1 Comment

My open-source sabbatical is ending in June (but before I will do a 45-day challenge) I would love to start early conversations about my next move (DMs are open.) Starting Monday, I will do a 45-day challenge to rebuild Fonoster from scratch. Some of my goals include: - Implement ASR and TTS from all major vendors - Double down on the Voice Autopilot - Improve observability and reporting - Migrate from Routr v1 to v2 - Bring SMS to the platform - Rebuild the frontend - Simplify DevOps As part of the challenge, I will post regularly and share milestones. Hit the 🔔 for updates. ~~~ ✍️ What new features would you like to see? ♻️ Reshare to your friends in the CPaaS industry

38

4 Comments

"Collaboration may sound straightforward, but it takes a lot of effort, communication, and focus to make it happen. And this need for teamwork is particularly true regarding making IoT technologies more seamless and reliable." Naama Zarfati, Augury’s Senior Product Manager for Connectivity, is sharing how she leads teams that align with each other and embrace best practices across the industry, allowing us to arrive at the next generation of IoT. Check it out here 👇 #IoT #Augury #IndustrialAI

8

Just when you think we are getting used to the steady improvements of the capabilities of the LLMs, especially the foundation models – this happens; Anthropic releases Claude 3.5 Sonnet. The week has already been buzzing with various benchmarks, tests and stories about Claude 3.5 Sonnet, but the one that I found stands out is the quality of code generation and the HumanEval benchmark. The HumanEval benchmark contains 164 programming challenges which are handcrafted with 7.7 tests per problem. Claude 3.5 Sonnet scores 92% at the HumanEval benchmark (0-shot)!! Also, another significant one is the GPQA 5-shot benchmark. GPQA is the Graduate level Google-Proof Q&A benchmark, which contains 448 multiple-choice challenging questions. Claude 3.5 Sonnet scored 67.2% !  Just for comparison – expert in-domain PhDs score 65%. This has led to update Alan���s conservative countdown to AGI at 75% !! One step closer to AGI ? Keeping the seatbelts tightly fastened. https://lnkd.in/ebzPfwMu

21

2 Comments

I remember learning category theory in undergrad and wondering: when would this ever be useful? Turns out it's very useful for program synthesis and software modeling -- thanks Kristopher for the great talk!

9

New Mindset Needed for Large Language Models - https://buff.ly/3XbN4vp #AI #ML #LLMs #genAI

Say you have mission critical parts, interesting new products to develop, and invest hugely in R&D in hardware, but you had to use the same horizontal procurement software that everyone else uses and was built 20-40 years ago. No automated RFP and invoice management, no supply chain risk scoring (vendor or part based), no dynamic assessment of critical needs and potential delays. That sounds... suboptimal. Forge (YC W24) gives hardware teams the tools to accelerate.

13

1 Comment

Interesting article that reminds me of the ego and arrogance that preceded the dot com bubble burst, and then the corporate convictions for monopolistic practices. For those of us with long memories and tenured careers, let’s prevent history repeating itself in the age of AI.

21

1 Comment

It’s exciting to witness how the hyperscalers are enabling a hyper-collaborative marketplace ecosystem that allows the multiplication of revenue opportunities. I see this happening in two ways 1️⃣ ISVs are being encouraged by hyperscalers to build with interoperability in mind 🔧 All three #hyperscalers have, over the course of the last few years, incrementally deepened their partnerships with on-page "competitors" like Salesforce and Snowflake to break data silos and leverage #GenAI on more comprehensive data sets. AWS, GCP and Azure are not just strengthening their own associations, but they are also actively encouraging collaboration between these ISVs to drive interoperability, aimed at the creation of joint solutions that significantly expand the universe of customers who can benefit. Snowflake’s most recent Polaris Catalog launch also reaffirms its commitment to interoperability with a wider library of industry players (beyond the three clouds) including hyperscaler partners like Salesforce and Confluent. That brings me to point 2…. 2️⃣ Hyperscalers are actively inviting the participation of channel partners to piece complementary solutions together 🤝 As ISVs focus on creating compatible offerings with each other, and with hyperscaler services, buyers have the opportunity to find end-to-end solutions for specific use cases right from the marketplace catalog. To do so, they’re relying on partners to identify, connect and manage these increasingly complex technologies. This presents a tremendous opportunity for channel partners to add value to these marketplace-led deals by helping piece solutions together into a comprehensive whole. Hyperscalers are also cognizant that the “channel is the glue between them and the lower enterprise, midmarket and SMB,” as I’ve hear my friend and founder of Ultimate Partner Vince Menzione put it, and are actively developing strategies to incorporate channels into marketplace deal workflows. There’s a clear signal to shift the perception of marketplaces from being potential threats to traditional partnership associations, to what they really are – growth opportunities. Channel partners who recognize this sooner rather than later, have biggest opportunity to capture net new business.

16

1 Comment

For folks working on Conversational AI - what is your network latency and jitter budget for great QoE? I assume good QoE would mean less than 400 ms of round trip delay and very consistent delay (jitter). Appreciate any insights! #conversationalAI #artificialintelligence Came to mind after seeing a demo and the mobile device was connected to the network via USB-C-based Ethernet - to factor out network delay.

18

1 Comment

Can’t agree more on this as we are seeing more early stage startups building more useful AI-native agents instead of copilots on existing workflow.

4

Challenging Traditional Investment with Vinod Khosla’s AI-Driven Vision Vinod Khosla, the founder of Sun Microsystems and my personal absolute investor idol, diverges sharply from traditional investment icons like Buffett and Munger. His approach? Embrace technological innovation and the power of learning from failures. While Buffett openly admits his gaps in tech understanding, as seen here: https://lnkd.in/dgbnWYpY, Khosla thrives in this uncertainty, advocating for a bold, fail-fast mentality that drives progress in the tech sector. >> Vinod’s strategies aren’t just about investing; they’re about redefining value creation. For more incredible insights, check out this talk by Vinod: https://lnkd.in/de6qchPi. Let’s explore how his fearless approach can inspire us to reshape our own investment philosophies. How are you integrating innovative thinking into your investment strategies? https://lnkd.in/de6qchPi #CreatingtheFuture

15

Key pain points of LLM evaluation well summarized by Andrew Ng: - Lack of testing dataset / expensive to curate manually - Difficult to create good metrics to capture nuances - Expensive to run eval (can easily cost ~$200 / day in his estimate) - Time-consuming to run - makes iteration slower If you've been reading our blog posts, you would know how Relari / continuous-eval are addressing these pain points!

24

Daniel Wilson and Reza Rassool Thanks for the opportunity to join the conversation, and more importantly to contribute to the critical work of Kwaai Personal AI -- turning AI into a meaningful benefit versus a threat in people's lives. (My comments start about 30 min in, after Reza's introduction to Kwaai))

3

2 Comments

I like this idea of "technical taste". It gets into where software engineering is as much creative art as it is science. Four great takeaways (IMHO) for software devs in this era in particular: 1. Aspiring engineers should cultivate a sense of curiosity, experiment with different tools and technologies, and embrace a mindset of continuous learning. 2. AI has the potential to streamline processes and enhance productivity for engineers, but it may also lead to disruptions in traditional software development workflows. 3. Developing technical taste and judgment is essential for making informed decisions about which technologies and approaches to pursue. 4. Collaboration and open-mindedness are key to leveraging the full potential of AI and staying ahead of technological advancements. I also really liked these two quotes from Sam Schillace: 1. "The right time to do something is when you have that feeling in the pit of your stomach that's like, 'oh, this is a great idea and it's going to suck to build because nothing's ready yet.'" 2. "Technical taste is like, 'how well have you consolidated that set of experiences and heuristics into judgment that you can apply accurately when you see new things?'" 3. "It may be the case that very small teams can do very large projects, or like we were talking about before, it may be the case we're just going to get really ambitious about what we try to do with the same size of teams, which is kind of where I would put my money." Great job on continuing to put out great episodes, Brett Berson and team First Round Capital! #ai #engineering #softwaredevelopment

5

Join the Sui Overflow hackathon and compete for a share of $5000 in Sui by integrating BlockEden.xyz's RPC and Indexer services. Learn how to participate, the prize distribution, and judging criteria in our latest blog post. Discover the benefits of using BlockEden.xyz and start building your innovative project today!

1