Explore Scalable Digital Solutions #5 - Get the BASICS right
2024 AWS Summit in Berlin is one of the most influential, original and thought-provoking annual tech event in Germany, across many industries from financial services to manufacturing and engineering.
Here you can discover how the latest advancements in generative AI, learn from organizations how they developer tooling, how to build foundational infrastructure allow you to build quickly with confidence, differentiate your business through your data, and deploy applications with availability, elasticity, cost, sustainability, and performance in mind, you can talk with an architect, play and learn, meet the AWS community in creative problem solving, at the historical, 150 years old Station Berlin, which used to be a train station from Berlin to Dresden, established in 1875.
My key takeaway is to get the BASICs right
Baseline your data foundation
Marketing buzzwords such as "Platform", "GenAI", can strengthen repetition Bias by simplifying evaluations of a concept by focusing on information that has repeated words, that might be quite distant from reality.
To truth unleash the power of AI, we need good data.
During the keynote, Mindy Ferguson, AWS Streaming & Messaging AWS, has emphasized the importance of comprehensive generative AI stack, with solid data governance, enabled with infrastructure that enables training and inference, apply tools to build application with FMs (functional models) and LLMs (Large Language Models).
High-quality data is the foundation of any successful AI endeavor, it's well worth investing time and resources to get it right. From a Data science perspective, there're different types of needs.
Algorithm at scale
An algorithm is a set of rules or a step-by-step procedure designed to perform a specific task or solve a particular problem. In AI, algorithms are the core mechanisms that drive the functionality of models and systems. Today, with the emergent of ChatGPT, we are seeking a paradigm shift in terms of the amount of parameters in foundation models and large language models.
For example, the first AI model was created back in 1957 with 3 parameters, Google's PaLM (Pathways Language Model) is one of the most advanced models and has 540 billion parameters, to generate original output for tasks like answering questions, translating languages, and completing sentences.
Today, Foundation models represent a high-level, generalized approach in AI, providing a broad platform for diverse applications. Developing a Foundation model involves training on vast datasets to create a versatile base. But there is not one that will rule all, often it requires a contextual linkage, e.g. applying Retrieval Augmented Generation (RAG).
It is the process of optimizing the output of a large language model, so it references an authoritative knowledge base outside of its training data sources before generating a response. RAG extends the already powerful capabilities of LLMs to specific domains or an organization's internal knowledge base, all without the need to retrain the model. It is a cost-effective approach to improving LLM output so it remains relevant, accurate, and useful in various contexts.
Strength with high resilience
The European Commission is determined to boost Europe’s competitiveness and innovation in financial services, and to be a global standard-setter. One aspect of the Commission’s drive for digital innovation in financial services is The Digital Operational Resilience Act (DORA):
It is a proposal for a pan-European framework on operational and cyber-resilience. The Act outlines improvements in information and communications technology (ICT) and security risk-management requirements, a harmonization regime for ICT incident reporting, development of a digital operational resilience testing framework, and an oversight framework for critical ICT third-party providers, in areas of:
领英推荐
Internal Developer Platform (IDP)
In recent years, BMW has been setting up a common internal development platform (IDP), CodeCraft, which is part of its overall “Software Factory.” During BMW’s development cycles, this software development platform executes at peak more than 140,000 continuous integration and continuous delivery (CI/CD) builds daily, using over 74,000 virtual CPUs. These builds can be very complex, relying on up to thousands of direct and transitive dependencies. Frequently, it takes several hours to generate artifacts up to several hundred GBs, causing two major challenges:
During another talk of IDP, AWS expert has shown best practices of how to build IDP, there is no one size fit all, however the underlying principles are universally applicable, by balancing autonomy and standards.
Capability building as priority
AWS has announced to invest 7.8 billion EUR in Brandenburg infrastructure, to create the capability of the future with European Sovereign Cloud.
How to serve the future if the organization doesn't have the right competence and capabilities? In fact more than half of the organizations are struggling with capability gaps in cloud talent, AWS is investing more than ever the training and certification program.
Amazon Web Services (AWS) has made significant commitments and efforts towards sustainability.
To summarize, by getting the following right,
the future is bright for builders, in fact, it's never been a better time to be a builder. The combination of advanced technologies, accessible resources, global connectivity, and supportive ecosystems creates a fertile ground for builders to innovate and thrive. the opportunities are vast and the tools at your disposal are more powerful than ever, with the dazzling speed of innovation opening doors to endless possibilities.
Source: AWS, Scaled Agile Inc.
Helping companies to adopt DevSecOps best practices!
6 个月Heard only good things from my colleagues that attended the summit as well. I guess, I need to put it on my agenda for next year. Thank you for sharing these insights!