Henry Weller

How we index timeseries data at Datadog with Rust and RocksDB https://lnkd.in/e8x_RUmz

56

3 Comments

If you want to learn about systems that power gen AI (and how Codeium does this in a novel way)- this presentation from my man Kevin is a must watch!

7

1 Comment

I’ve gotten some questions regarding 1. Which LLMs does Salesforce support? and 2. How do I take advantage of generative AI tools while keeping my data and customer’s data safe? To briefly answer this question; currently, the Models API supports large language models (LLMs) from multiple providers, such as Amazon Bedrock, Azure OpenAI, OpenAI, and Vertex AI from Google! You can find the full list here: https://lnkd.in/eH-gTDS6 and in fact.. Salesforce has also gone above and beyond launching the world’s first LLM benchmark for CRM to assess the efficacy of generative AI models for business applications: https://lnkd.in/eG5bDjWs. This benchmark will compare accuracy, cost, speed as well as trust and safety... and it’s available to every user! For the second question, Salesforce has the Einstein Trust Layer in place which ensures the security of generative AI through data and privacy controls that are seamlessly integrated into the end-user experience. This layer combines Data Masking, Prompt Defense, Dynamic Grounding, and a ZERO-Retention Policy which ensures your data will NOT be stored outside of Salesforce, while delivering a delightful experience to customers. To find out more on Einstein Trust Layer, complete this trail! https://lnkd.in/eXJPpcWf   #Einstein #AI #Salesforce

12

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

Tiered storage for Kafka is a classic tarpit idea. It makes all the sense in the world, but it doesn't work in practice. Check out our latest blog post to learn why. https://lnkd.in/eG9BJ95t

107

8 Comments

Custom LLMs are starting to emerge for select workloads that require very high computational power and deep customization for particular use cases, such as Healthcare and FinServ. We're building a partnership with Ginkgo Bioworks to help them accelerate drug discovery and drive better patient outcomes. The models are huge -- trained on billions of proprietary Ginkgo protein sequences -- and Ginkgo is now making these models publicly available for researchers and developers via a programmer-friendly API. There's a nice article in Forbes out today which lifts the curtain a bit of what's going on and where we're headed. https://lnkd.in/gTUU45v5

11

With the announcement of S3-native-streams (Freight clusters), here is a commentary on Confluent strategy regarding object storage, streaming and an open data architecture. https://lnkd.in/dTg6DKVy

49

6 Comments

More details about the mighty phi-3 family of SLMs. Such a key development as GenAI Dev becomes more about always optimizing and orchestrating with LLMs routing to LLMs and SLMs continuously to balance complexity, latency, cost and more! https://lnkd.in/gGQYqNKt

9

Missed AICamp? Developers, catch the replay of our webinar featuring Google Cloud and Redis! Learn how to: - Build AI agents with Vertex AI to enhance your apps - Leverage Redis vector search for powerful RAG & LLM applications Whether you're new to AI or looking to level up, these talks are packed with insights. Watch the recording: https://lnkd.in/d6vABUHd https://google.smh.re/4TvS

2

Great article by Roi Yarden Building Data pipelines comes with its set of challenges. One of the most important decisions is to understand what patterns suit the business evolving needs and allow for a system to scale. This can prevent very costly mistakes.

8

In this demonstration, Tom Scott tackled a real-world scenario where a retailer's transaction data was producing erroneous results due to an unusually large exchange rate for the Zimbabwean Dollar. This anomaly caused transactions with this currency to be recorded with a zero value, effectively making them invalid and impacting financial reports and operations. This anomaly caused the transaction amounts to reduce to zero, disrupting operations and creating confusion. By leveraging #indexing, we swiftly identified the root cause of the problem—the incorrect exchange rate—allowing us to rectify the issue promptly. This example highlights how indexing significantly accelerates problem-solving, enabling teams to quickly diagnose and address data anomalies without the need for unnecessary data movement and the complexity of #ETL pipelines. #Kafka #DataStreaming #Indexing #DataAnalytics #BigData #RealTimeData #ETL #DataManagement #StreamProcessing

9

When we initially built Conduktor, as Kafka engineers, our focus was to provide other developers with the most helpful tool—removing all the complexity of the CLI— and making it accessible in an elegant, simple, but powerful UI. The more Conduktor grew, the more we hit some limitations inherent to UIs: applications don't talk to a UI, they don't talk to an HTTP API, they talk to Kafka directly. Somehow related, in the most mature tech organizations, everything revolves around GitOps: it provides the core experience with all its constraints, but at least, you have control! It is often a beautiful dream, because lacking flexibility and causing much friction for the teams who need to use it. This leads to operational burdens and complexity evolving at a fast pace to answer the developers and ops needs. At Conduktor, we wanted something more flexible and easy to deploy no matter your tech maturity, because time is precious for everyone. Everyone loves simple stuff. I do. It's easy to build complex things and harder to build simple things. We enhanced Conduktor by connecting it directly to the Kafka applications, too, offering many features beyond standard Kafka capabilities (preventing misconfigurations, header mutations, easy encryption, multi-tenancy, data quality constraints, etc.) and paving the way to unify the user and application experience. While having a beautiful engine purring is great, the real value comes in making the technology accessible and collaborative for all teams, helping accelerate business objectives. Everyone, including organizations, needs guidance and a path to grow. Data management is complex to get more or less right (talk to your CDO), and there are so many paths ahead. Streaming data is complicating the landscape even more. This led us to transform Conduktor again into a streaming framework atop your data ecosystem, designed to standardize workflows and enhance team collaboration. This is the essence of Conduktor 2.0. TLDR: Get started and let me know https://www.conduktor.io TLDR2: Read more here about our new free tier and our Kafka features here https://v2.conduktor.io/

3

Very useful insights for folks who are starting to dig further and further into the realm of LLM. An easier GUI approach for fine tuning LLMs.

9

Truth. This blog articulates the shortcomings of most tiered storage solutions in delivering predictable performance at scale. Without some rebuild that harnesses cloud primitives like AWS S3 and GCP GCS, mere incremental adjustments to existing products won't adequately serve users. From a product manager's perspective, sometimes incremental adjustments isn't a path to sustainable product adoption and success.

10

The future is Fine-Tuned! Join me and Daryl Martis at 10:30 PT today to learn about how to easily fine-tune a small open source LLMs in Salesforce to outperform the large GPT 4.0 type models in terms of quality, speed and cost. Small models, big impact. #df24 #finetuning

64

Awesome to see the Meta #Llama 3.2 models on #Azure #AI Model Catalog through managed compute inferencing #MaaS and soon on serverless API deployments while MACC Eligible providing customers with Enhanced Security and Compliance, Content Safety Integration, Simplified Assessment of LLM flows, Client integration, and Simplified Deployment and Inference. https://lnkd.in/g4Zpb3Dm

9

1 Comment