Extending Postgres With Sync Engine #SfTechWeek

​Git gave us diffs. React gave us virtual DOM diffing. Kafka gave us event streams. Every layer of the stack learned to ship only what changed, except the query you just ran against Postgres.

​We built StreamGres to fix that. It sits next to Postgres and lets clients subscribe to a query once. From then on, they receive only the rows that entered, left or changed in their results, as soon as they're committed. This talk covers how we built it and how we use it in production.

​How we built it:

​⚡ Routing each write only to the queries it affects, no matter how many are registered

​🧩 Sharing data across thousands of subscriptions, so overlapping queries are served fast

​How we use it:

​StreamGres powers real-time collaboration in Xyne: an open-source, self-hosted AI Org OS for people and agents.

​Bonus: XOR We’re demoing XOR, our open-weight decision model. Give it text, images, or video along with a set of multiple-choice questions, and for each question it returns the probability of every option. Model: huggingface.co/juspay/xor

​If you're using a sync engine or local-first ideas in your application, apply here to showcase your product: Open Google Form

​Live demos, then Q&A with the engineers who built them. This event is a part of #SFTechWeek—a week of events hosted by VCs and startups to bring together the tech ecosystem. Learn more at www.tech-week.com

​Arun Raj, Architect, Juspay · Saket Raman, R&D, Juspay · Aniket Agrawal, R&D, Juspay