While I heard some people referring to this event as “Spring Break for Geeks,” this event felt like the center of the modern data stack universe. MDS companies were everywhere; both well-known big players but also data startups trying to make a name. As a data meeting, buzzwords were well-represented; Generative AI, the Semantic Layer, data observability—and did we say AI?
This was the first time I’ve attended Data Council and was impressed by the caliber of attendees, the approachability of the venue—and the sheer number of happy hours! But make no mistake, content was King.
Keynote talks were of exceptional quality ranging from DJ Patil, former U.S. Chief Data Scientist and inventor of the term “Data Scientist,” to Jordan Tigani’s “Big Data is Dead” talk. There was so much to take in over the three-day event, but these are my top 3 takeaways:
AI is now! Many AI experts were present, but even those not in the field were trying to figure out the implications. The industry is driving for cheaper, faster, and smaller models, and as LLMs become ubiquitous it seems training data and UX could be an important differentiator. Getting the mode of interaction right to make AI feel natural will be critical.
The Modern Data Stack is Spaghetti with Tomato Sauce kinda mess: Many talks and startups highlighted the duality between problems the MDS has solved and the problems it’s created. The fundamentals matter more than ever: ROI, costs, data quality, and observability. Bundling is a trend to watch out for.
WTF is the semantic layer? This buzzword was everywhere but it’s a critical part of making data accessible for non-data stakeholders. The industry continues its trend in democratizing data and fueling the need for data-driven insights. Augmenting SQL and AI-based querying are making their way into the data warehouse.
Here's a video of my talk from Data Council 2023, on the topic of scaling experimentation to 20 billion users. Enjoy! :)
Thanks to our support team, our customers can feel like Statsig is a part of their org and not just a software vendor. We want our customers to know that we're here for them.
Migrating experimentation platforms is a chance to cleanse tech debt, streamline workflows, define ownership, promote democratization of testing, educate teams, and more.
Calculating the right sample size means balancing the level of precision desired, the anticipated effect size, the statistical power of the experiment, and more.
The term 'recency bias' has been all over the statistics and data analysis world, stealthily skewing our interpretation of patterns and trends.
A lot has changed in the past year. New hires, new products, and a new office (or two!) GB Lee tells the tale alongside pictures and illustrations:
A deep dive into CUPED: Why it was invented, how it works, and how to use CUPED to run experiments faster and with less bias.