Hi everyone!!!! Some conferences you go to because you're invited. Some you go to because you've been meaning to for years. TiDB SCaiLE is the second kind for me.
I first noticed it in 2023, when it was still called HTAP Summit. I planned to go that year. And the next. And the one after that, when it became TiDB SCaiLE. Every year, something else won the calendar.
This year it doesn't. On October 15, I'll be at the Computer History Museum in Mountain View, covering TiDB SCaiLE 2026 in person for The Ravit Show.
Why this one
Agentic AI is changing what a database has to do.
When one user action spawns thousands of agents, each one needs its own context, its own memory and fast access to data. Vectors, transactions and analytics all have to sit together. Most stacks today stitch that together across several systems, and the engineering teams keeping it running are paying for it.
SCaiLE is where the people dealing with that problem talk openly about it. The call for talks asked for real numbers, real tradeoffs and real production scars. That's the kind of session I want to be in.
Who's speaking
This list is why I booked my trip:
- Vinod Kumar, VP and Head of Engineering, Atlassian
- John Grass, Senior Manager, Storage Foundations, Pinterest
- Tailin Lyu, Senior Software Engineer, Pinterest
- Nilanjan Chatterjee, Sr. Staff Data Architect, AMD
- Yijia Zhang, Head of AI and Platform, Plaud
- Madhav Ramesh, Software Engineer, Databricks
- Lang Fei, Global Tech Lead, AWS
They're joined by PingCAP Co-Founder and CEO Max Liu, Chief Architect Sunny Bains, and builders from Harmonic, Braintrust, OpenHands, Dify and Memori Labs.
What I'll be tracking
The program runs on 2 tracks.
Infinite Agents: designing for workloads where software is the primary user. Think per-agent isolation, memory layers, vector retrieval, and creating databases in under a second.
Infinite Data: running transactional and analytical workloads at global scale. Zero-downtime migrations, multi-region resilience, and what held up under load.
I'll be asking speakers about the parts that usually don't make the slides. What broke, what they'd change, and where the costs showed up.
Join me
If you build or run data infrastructure, this is a day well spent. It's one day, in one room, with people who've done the work.
If you're coming, reply and let me know. I'd love to meet in person.
See you in Mountain View,
Ravit
