Purpose: To bring together ML researchers and ML practitioners for open discussions and projects on problems in production ML.

This interest group is run as a collaboration between MLCollective and Full Stack Deep Learning.

MLC is a global, open, distributed ML research lab. FSDL ****is a MOOC and community focused on DL systems “across the stack”: from conceptualization and data collection to model selection, continual monitoring, and product design.

For questions, ping @charles_irl on Twitter or Discord.

<aside> 💬 Join the MLC Discord here! We’re in #production-ml under “Interest Groups”. Make sure to sign up for notifications by reacting 🥞 to this message after you’ve joined the server.

</aside>

Logistics

We have semi-synchronous discussions on the MLC Discord on topics from research papers relevant to production ML to examples of ML-powered technologies to tools for doing ML in production.

We meet synchronously and virtually for discussions weekly on Mondays from 9am to 9:30am Pacific. Subscribe to the MLC Events Calendar, embedded below, for up-to-date information, including meeting links.

<aside> 📆 During events, you can join us live here.

</aside>

We’re currently experimenting with new forms for our meetings, so join the Discord for the latest on what we’re up to.

Event Series

ProdML Tooling Sessions

As our current meeting format, we’re taking inspiration from Full Stack Deep Learning's "Tooling Tuesdays" and sharing tips, tricks, and learning experiences with tools for production ML.

Here’s some tools we might cover: PyTorch Lightning, Gradio, Hugging Face, Weights & Biases, DVC, LabelStudio, Great Expectations, Snorkel, TorchScript, FFCV, GitHub Actions, Dagster, Cog, MosaicML Composer, kubernetes

<aside> ✋ We're looking for volunteers to present at future tooling sessions. Pick a tool you like or you want to learn and share it with the group. You'll want 15-20 minutes of demo content (e.g. walking through an example from the tool’s docs), then we'll do ~10 minutes of discussion. DM @charles_irl on Twitter or Discord to schedule. You don’t need to be an expert in the tool, just willing to give it a try and share how it went!

</aside>

Designing ML Systems Reading Group

As our first event series, from March to May 2022, we read Designing ML Systems by Chip Huyen.

The meetings were recorded and uploaded to YouTube. The videos are also embedded below.

https://www.youtube.com/watch?v=K-SuQbmttbg&list=PLfeYlJzwvDN3R_xBwA8pp0cLOzxndlC4X&index=1&ab_channel=MLCollective

Links