Date: Wed, 21 Sep 2022 16:51:00 +0000
<h1><span style="font-weight: 400;">Vertex AI Experiments with Ivan Nardini and Karthik Ramachandran</span></h1> <p><span style="font-weight: 400;">Hosts</span> <a href="https://twitter.com/asrivas_dev"><span style="font-weight: 400;">Anu Srivastava</span></a> <span style="font-weight: 400;">and Nikita Namjoshi are joined by guests</span> <a href="https://twitter.com/ilnardo92"><span style="font-weight: 400;">Ivan Nardini</span></a> <span style="font-weight: 400;">and</span> <a href="https://twitter.com/karkumar"><span style="font-weight: 400;">Karthik Ramachandran</span></a> <span style="font-weight: 400;">in a conversation about Vertex AI Experiments this week on the podcast. Vertex AI Experiments allows for easy, thorough ML experimentation and analysis of ML strategies.</span></p> <p><span style="font-weight: 400;">Our guests start the show with a brief introduction to Vertex AI and go on to help us understand where Experiments fits in. Because building ML models takes trial and error as we figure out what architecture and data management will work best, Experiments is a handy tool that helps developers try different variations. With extensive tracking capabilities and analysis tools, developers can see what is working, what’s not, and get ideas for other things to try. Ivan tells us about the two concepts to keep in mind before using Experiments: runs, which are training configurations, and experiments, adjustments you make as you look for the best solution.</span></p> <p><span style="font-weight: 400;">Vertex ML Metadata, a managed ML metadata tool, helps analyze Experiment runs in a graph, Ivan tells us. He takes us through an example ML model build and training using Vertex AI Experiments and other tools. He and Karthik also elaborate on the relationship between Vertex AI Experiments and Pipelines. We talk about the future of AI, including the foundational model, and some cool examples of what’s happening in the real world with Vertex AI Experiments.</span></p> <h5><strong>Ivan Nardini</strong></h5> <p><a href="https://twitter.com/ilnardo92"><span style="font-weight: 400;">Ivan Nardini</span></a> <span style="font-weight: 400;">is a customer engineer specialized in ML and passionate about Developer Advocacy and MLE. He is currently collaborating and enabling Data Science developers and practitioners to define and implement MLOps on Vertex AI. He is an active contributor in Google Cloud.</span></p> <h5><strong>Karthik Ramachandran</strong></h5> <p><a href="https://twitter.com/karkumar"><span style="font-weight: 400;">Karthik Ramachandran</span></a> <span style="font-weight: 400;">is a Product Managed on the VertexAI team. He’s been focused on developing MLOps tools like Vertex Pipelines and Experiments.</span></p> <h5><strong>Cool things of the week</strong></h5> <ul> <li style="font-weight: 400;"><span style="font-weight: 400;">Expanding the Google Cloud Ready - Sustainability initiative with 12 new partners</span> <a href="https://cloud.google.com/blog/topics/sustainability/how-partners-benefit-from-google-cloud-ready-sustainability"> <span style="font-weight: 400;">blog</span></a></li> <li style="font-weight: 400;"><span style="font-weight: 400;">Large Language Models and how they are used with Natural Language Understanding.</span> <a href="https://arxiv.org/pdf/2208.11857.pdf"><span style="font-weight: 400;">pdf</span></a></li> </ul> <h5><strong>Interview</strong></h5> <ul> <li style="font-weight: 400;"><span style="font-weight: 400;">Vertex AI</span> <a href="https://cloud.google.com/vertex-ai"><span style="font-weight: 400;">site</span></a></li> <li style="font-weight: 400;"><span style="font-weight: 400;">Vertex AI Experiments</span> <a href="https://cloud.google.com/vertex-ai/docs/experiments/intro-vertex-ai-experiments"> <span style="font-weight: 400;">docs</span></a></li> <li style="font-weight: 400;"><span style="font-weight: 400;">Vertex AI SDK for Python</span> <a href="https://cloud.google.com/vertex-ai/docs/start/client-libraries#python"> <span style="font-weight: 400;">docs</span></a></li> <li style="font-weight: 400;"><span style="font-weight: 400;">Vertex ML Metedata</span> <a href="https://cloud.google.com/vertex-ai/docs/ml-metadata"><span style="font-weight: 400;">docs</span></a></li> <li style="font-weight: 400;"><span style="font-weight: 400;">Vertex AI Pipelines</span> <a href="https://cloud.google.com/vertex-ai/docs/pipelines"><span style="font-weight: 400;">docs</span></a></li> <li style="font-weight: 400;"><span style="font-weight: 400;">Vertex AI Workbench</span> <a href="https://cloud.google.com/vertex-ai/docs/workbench"><span style="font-weight: 400;">docs</span></a></li> <li style="font-weight: 400;"><span style="font-weight: 400;">Vertex AI Tensorboard</span> <a href="https://cloud.google.com/vertex-ai/docs/experiments/tensorboard-overview"> <span style="font-weight: 400;">docs</span></a></li> <li style="font-weight: 400;"><span style="font-weight: 400;">Track, compare, manage experiments with Vertex AI Experiments</span> <a href="https://cloud.google.com/blog/topics/developers-practitioners/track-compare-manage-experiments-vertex-ai-experiments"> <span style="font-weight: 400;">blog</span></a></li> <li style="font-weight: 400;"><span style="font-weight: 400;">Vertex AI Experiments Notebooks</span> <a href="https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/official/experiments"> <span style="font-weight: 400;">site</span></a></li> </ul> <h5><strong>What’s something cool you’re working on?</strong></h5> <p><span style="font-weight: 400;">Anu is working on demos for</span> <a href="https://cloud.withgoogle.com/next"><span style="font-weight: 400;">Next</span></a><span style="font-weight: 400;">.</span></p> <p><span style="font-weight: 400;">Nikita is testing new features for</span> <a href="https://cloud.google.com/vertex-ai"><span style="font-weight: 400;"> Vertex AI</span></a><span style="font-weight: 400;">.</span></p> <h5><strong>Hosts</strong></h5> <p><span style="font-weight: 400;">Nikita and Anu Srivastava</span></p>