PyData Seattle 2023 In Review
The Union.ai team kept busy last week at PyData Seattle 2023. The annual conference brings together data scientists, engineers, business leaders and Python enthusiasts to explore the latest trends in data tools and technology.
We were thrilled to participate in the event as a Diamond sponsor and engage with attendees through talks and tutorials, conversations at the booth, and over some drinks at our happy hour. Read on to get the full scoop on all things Union @ PyData!
Talks and Tutorials
PyData Seattle talks and tutorials spanned three days at the Microsoft Conference Center in Redmond, Washington. This year’s key themes included machine learning, LLMs and data visualization.
Union Engineering Manager Eduardo Apolinario kicked off Union’s contributions Wednesday, April 26. His tutorial, “Flyte: Robust and End-to-End Cloud Native Machine Learning & Data Processing Platform,” provided attendees an overview of how they can use Union’s open-source Flyte orchestration engine to scale and manage machine learning workflows in production.
On Thursday, April 27, Flyte community member Shivay Lamba turned the spotlight on UnionML — Union’s open-source Python framework — in his tutorial, “Building Machine Learning Microservices & MLOps using UnionML.” Shivay introduced some of the core concepts of building and deploying machine learning microservices across environments with a standardized and uniform interface.
On Friday, April 28 (the last day of the conference), Union.ai CEO Ketan Umare and founding engineer Yee Tong presented “The Python Data Ecosystem: Navigating a fragmented landscape.” This talk covered the current challenges to organizations building scalable ML workflows. Ketan and Yee also described how Flyte could solve some of those as a data-aware and highly extensible orchestrator that can be used by both data scientists and ML engineers. The duo showcased two demos: The first was a Flyte workflow that tested performance across multiple models on the same dataset.The second demo previewed some of the latest work from the Union team, with a Flyte workflow for fine-tuning LLMs! Recordings of the conference sessions will be posted on the PyData TV YouTube channel in the coming weeks, so stay tuned if you weren’t able to attend in-person this year.
Insights from the booth
In the conference hall, attendees could stop by Union’s booth and speak to our engineers, product experts, Flyte community managers and dev advocates. We talked to data scientists, business analysts, ML engineers and a ton of Python enthusiasts. Conversations were heavily focused on workflow orchestration for both data and ML use cases in production (think ETL pipelines and machine learning workflows) and where our products fit in the general data-tooling ecosystem. We also fielded a few questions about life at Union (because we are hiring)!
Happy Hour at Locust Cider
In addition to our presence on-site at the conference, we moved off-site on Thursday night for a happy hour at Locust Cider. We were so excited to provide PyData attendees and Flyte community members alike an opportunity to socialize and network.
Cheers to the Python community
We want to thank NumFocus for organizing this event and providing all of us data and Python enthusiasts a space to share our knowledge and insights across the community. And we’re looking forward to the next PyData!