llms Querying a structured data with an LLM and Llamaindex This is a deep dive into the exact logic I used to get the information requested by the user from the structured data store in the previous post.
aws Amazon Bedrock Agent with an action group to query structured data Here is a high-level waltkhough of how to build an Amazon Bedrock agent with an action group to query a structured data store.
data analytics LLM for data analytics: 3 main text-to-sql architecture patterns Althrough, already today SQL AI agents can be effective for very limited and well-defined use-cases, text to sql has a long way to go to replace data analysts and similar professionals. In the below video I describe the three main architecture patterns when it comes to building a text to
aws ML pipeline on AWS using SageMaker Recently I had to consult a client on moving an on-prem recommender system to AWS. Naturally they were interested in using the SageMaker offering for building the entire train-validate-deploy pipeline. To refresh my ML pipelining knowledge I did this quick end-to-end tutorial of bulding an ML pipeline with Amazon SageMaker.
aws AWS Bedrock agent with knowledge bases primer A Bedrock Agent is one of the central offerings of the AWS Bedrock service. It is essentially an abstraction around a bundle of one of the Foundation Models (FMs) and a selection of additional tools: be it accessing a knowledge base or making use of any external functionality via a
data strategy Data strategy: how to get started An effective data strategy starts with the business and its key drivers. These drivers should be translated into overarching goals (strategic imperatives), which can then be broken down into actionable steps (business objectives). Only after these steps have been identified should the decision be made on how technology and data
data strategy The only data maturity scale that matters It is decidedly useless to know that you score three out of five in data accessibility and cleanliness based on one scale or another. Such ranking tells you nothing about your ability to achieve your business goals with data, which is a core idea of such a lofty concept as
data strategy Data strategy: why should you care? The idea that has been around for so long and was marketed so badly is experiencing a renaissance. A number of shifting circumstances invite executives to completely rethink the role of data in the organisation. From a bureaucratic necessity you need to fulfil for the upcoming audit—to a major
data strategy Data strategy in your company’s hierarchy of needs Your shareholders and your boss have been mulling over the idea for some time, even though subconsciously. So you can really make their and your lives easier by putting it on the table. Shedding light onto something that has been lurking in the background trying to come through is very
large language models LLM fine-tuning roadmap 2024 As with many life endeavours, you will get a head-start by having crystal-clear understanding of why you are fine tuning. Any foundation model out there is already pretty smart. It has been trained on vast amounts of text and maybe all you need to do is to give it a
data products On data products I like the idea. It sounds like a zero to one story. An act of creation. Compared to (long-sight), meh, dashboards. As tricky as it might be to define it, I will give it a go. On one end of the spectrum, there is an idea that in our age,