Andrew van Beek’s Post

Feature Friday: Sneak Peak at Internal Conversational Docs We're excited to share a sneak peek at a new capability we're working on - allowing you to have natural conversations with your documentation using the power of vector databases. Right now we already support this use-case for your own external documentation sites, but now we are looking towards bringing this capability to help onboard your engineers to your own codebases. Soon, you'll be able to simply ask Dev-Docs a question like "Is there a function that does X in this codebase?". Our platform will then search through your dev-docs markdown files, find relevant information, and generate a human-friendly answer summarizing the results. The really cool part? You will have the option to run the vector database locally on your machine(such as this demo) or sync it to our cloud service. It's still early days, but we can't wait to bring this conversational documentation experience to all of you. Check out the loom video below: https://lnkd.in/gaPxKEdW #DevDocs #ProductUpdates #DeveloperExperience #VectorDatabase #FeatureFriday

Just finished week 1 of mlops-zoomcamp presented by DataTalksClub and Alexey Grigorev. It is a kickstart to an amazing journey, and what I've learned so far is: 1. What is mlops? 2. Why do we need mlops? 3. How to lanuch EC2 instance on AWS 4. Prepare dev environment using Docker, docker-compose, EC2, Vscode, and jupyter notebook 5. MLOps maturity model 6. How to train and test data using NY-Taxi dataset Looking forward to start next week Week1 Assignment: https://lnkd.in/da_C_T-9

I'm excited to present my newest technical write-up, which concentrates on leveraging Amazon Bedrock's RetrieveAndGenerate API to reuse knowledge base sessions effectively. This article offers a detailed, step-by-step walkthrough on how to reuse a knowledge base session multiple times, ensuring optimal resource utilization and efficiency. For professionals currently working with Amazon Bedrock, this article provides valuable insights to further maximize the potential of its advanced features, enabling you to enhance your existing workflows and processes. Access the full article here: https://lnkd.in/eQVEuwPf Feel free to read through the content and share your thoughts, experiences, or any questions you may have. 🤝 #BigData #AmazonBedrock #CloudComputing #GenAI #AWS

Check how easy is to connect GPT to APIs in this webinar video which Zdenek Nemec and Martyn Davies recorded today.

At 3pm WEST, check how to add voice interface in your #outsystems applications. In this session will have Joao Salvado from #microsoft to share all the possibilities with Azure Speech services. https://lnkd.in/dNezNMej

🚀 Unlock the Power of Open-Source Vision-Language Models with PaliGemma from Google, Seamlessly deploy on our Inferless serverless platform! 🌟 ⚡ Experience Superior Performance with PaliGemma on Inferless with Average inference time of 0.86 seconds and average cold start time of 9.26 seconds ⚡ ️ 🌟 Embark on your journey to harness the power of PaliGemma by following our comprehensive tutorial. 🔗 Dive into the Tutorial: https://lnkd.in/gs5aR9jP

Week 2 of #LLMZoomcamp was all about Open-Source LLMs ✨ . We started with an introduction to open-source LLMs and explored using a GPU in Saturn Cloud 🌐. Some of the open-source LLMs we got our hands on included Google FLAN-T5, Microsoft's Phi-3-mini-128k-instruct, and MistralAI's Mistral-7B-v0.1 🧠. We also discussed where to find these models. Leaderboards, Google, and ChatGPT are all great resources for discovering open-source LLMs. Interestingly, we explored how to run an LLM without a GPU using Ollama 🖥️. Another exciting topic was building a UI for RAG using Streamlit 📊. If you're curious, you can check out my homework repository :https://lnkd.in/efTBCAZF. Looking forward to what week 3 has in store! Big thanks to Alexey Grigorev for this amazing course! 🙌

From Webhook to Log and Log to Metric: OpenTelemetry Magic! #observability #cloud #native #containers #metrics #logs #monitoring #otel

When a change needs to happen in an enterprise, I see the actions for change completed by however many teams are in the enterprise. By change, I mean switching to a new flavor Kubernetes, a new Cloud, or a new tool. After reading Team Topologies, I realized I worked on a different pattern. Most enterprises have feature teams, platform teams (in this case, runs infrastructure), and maybe a few special teams(AI, Big Data). I worked on a team that talked with the Platform team and created reference architectures, office hours, and Golden Paths for all the feature teams. An Enablement team. Take a gander at the concepts below and make enterprise changes at scale. https://lnkd.in/gdqfbTR6 https://lnkd.in/gxmWBsz7 #teamtopologies #devops

🚀 Elevate Your Time-Series Forecasting with TimesFM from Google, Seamlessly Deploy on Our Inferless Serverless Platform! 🌟 ⚡ Achieve Exceptional Performance with TimesFM on Inferless, boasting an average inference time of 0.12 seconds and an average cold start time of 35.84 seconds ⚡   🌟  Begin your journey to leverage the capabilities of TimesFM by exploring our detailed tutorial. 🔗  Access the Tutorial: https://lnkd.in/gRHmUpJN