New Show Hacker News story: Show HN: Learn ML and AI infrastructure in the browser
Show HN: Learn ML and AI infrastructure in the browser
7 by vtuulos | 0 comments on Hacker News.
We are the developers of an open-source package Metaflow that we started at Netflix. Metaflow provides a human-friendly interface to the full stack of ML infrastructure, including data access, compute, workflow orchestration, and versioning. It is used by hundreds of companies across industries. Over the past years, we have seen that there are two major stumbling blocks for folks who want to learn to build real-world ML applications: 1) Setting up the full infrastructure stack in the cloud costs time and money. The investment is worth it once you know what you want to do, but that's not always the case in the beginning when you just want to explore. 2) While many excellent ML/AI tutorials exist publicly, there aren't that many hands-on, interactive environments showing how to map models to real-world workflows, running on real infrastructure. This Show HN release is about our attempt to address (1) and (2): We host free, fully-featured sandbox environments that include a private Kubernetes cluster, metadata service and UI, and a workflow orchestrator backed by Argo Workflows (1). If you like what you see, you can set up a similar environment in your cloud account using our open-source Terraform templates. In addition, leaning on industry experts, we have created a number of tutorials covering NLP, computer vision, OpenAI Whisper, and recommendation systems (2). Hopefully you'll find these resources useful. Let us know what you think!
7 by vtuulos | 0 comments on Hacker News.
We are the developers of an open-source package Metaflow that we started at Netflix. Metaflow provides a human-friendly interface to the full stack of ML infrastructure, including data access, compute, workflow orchestration, and versioning. It is used by hundreds of companies across industries. Over the past years, we have seen that there are two major stumbling blocks for folks who want to learn to build real-world ML applications: 1) Setting up the full infrastructure stack in the cloud costs time and money. The investment is worth it once you know what you want to do, but that's not always the case in the beginning when you just want to explore. 2) While many excellent ML/AI tutorials exist publicly, there aren't that many hands-on, interactive environments showing how to map models to real-world workflows, running on real infrastructure. This Show HN release is about our attempt to address (1) and (2): We host free, fully-featured sandbox environments that include a private Kubernetes cluster, metadata service and UI, and a workflow orchestrator backed by Argo Workflows (1). If you like what you see, you can set up a similar environment in your cloud account using our open-source Terraform templates. In addition, leaning on industry experts, we have created a number of tutorials covering NLP, computer vision, OpenAI Whisper, and recommendation systems (2). Hopefully you'll find these resources useful. Let us know what you think!
Comments
Post a Comment