Qian Li
@qianl_cs
Co-founder @DBOS_Inc • CS PhD @Stanford ❤️Database+Architecture+Systems Co-organizing https://southbaysystems.xyz/ Bluesky: https://bsky.app/profile/qianli.dev
Timeouts are essential for building efficient and resilient systems. They help prevent systems from waiting indefinitely and free up resources while maintaining responsiveness under heavy load. For example, suppose your server must finish a task within 30 minutes, but some…

Just put together a fun new demo: an AI agent that researches Hacker News’s opinion of a topic. The agent starts with a topic, finds relevant stories and comments, analyzes them, then chooses related topics to explore until it has enough data. Then, it synthesizes everything…
The most rewarding part of building an open-source project is seeing a community grow around it. That’s why I love our monthly user groups, to meet the community and learn what people are building with DBOS. Tomorrow, we’ll hear a presentation about building scalable CI/CD on…
Join us in 2 days! The DBOS user group meetup features a deep-dive from Ontologize: learn how they train thousands of students every day on complex enterprise systems by deploying isolated software stacks at massive scale, reliably with DBOS. Plus, DBOS Go is picking up steam:…
Let's Go! 📣 Join us next week for the July DBOS User Group Meetup. We're unveiling an early preview for DBOS Go, one of our most requested languages! We'd love your early feedback as we explore this new direction. We're also excited to welcome the engineering team from…
In most video games, you can save your progress so that if you die, you don’t have to restart from the beginning. What if you could do the same in your programs? Save their progress so that if they fail, they can recover from where they left off? That’s essentially what durable…
Hope to save ATC and help build an even better systems community 🙏
A petition to SIGOPS to adopt the USENIX Annual Technical Conference (ATC) and retain its steering committee docs.google.com/document/d/1wK… (not sure whether it can be done by SIGOPS alone, but it's great to let the voice be heard)
Here’s a fun demo: an online storefront built on durable workflows. You can crash it as many times as you want, even in the middle of critical operations like payment processing, and it always recovers from exactly where it left off, as if nothing happened. This works because…
Let's Go! 📣 Join us next week for the July DBOS User Group Meetup. We're unveiling an early preview for DBOS Go, one of our most requested languages! We'd love your early feedback as we explore this new direction. We're also excited to welcome the engineering team from…
From talking to engineering teams, I think one of the toughest parts of operating an AI agent is observability. Debugging a conventional program is hard enough, but AI agents are even more difficult to reason about because they take actions based on unpredictable, sometimes…
If you have a space in South Bay (Palo Alto, Mountain View, Sunnyvale, Santa Clara, San Jose, etc), please reach out to us!
To South Bay folks: South Bay Systems is currently limited by finding venues willing to host. If you know of somewhere that can host 60+ people and is happy getting a few minutes for a quick "here's who we are, what we do, and who to talk to for hiring", please let me know!
When you're training thousands of students daily on complex enterprise software, you need reliable infrastructure. Ontologize, who provides hands-on training for clients including PwC, Conagra, and the US Air Force, faced a major challenge: seamlessly and automatically deploying…
New Screaming in the Cloud episode! Corey Quinn and our CEO @jedberg "chat about Jeremy's "build for three" rule, a plan for scale without going crazy, why he set Reddit's servers to Arizona time to dodge daylight saving time, and how DBOS makes your app as tough as your data."…

Excited to present at #P99CONF!
Most systems rely on external orchestration because it's well-established. But it introduces overhead, write amplification, and operational complexity. @qianl_cs will explore an alternative, lightweight library-based engine at #P99CONF. ow.ly/AFS250Wpegh #ScyllaDB
Happy weekend! I upgraded and "modernized" our cloud subscription backend to ESM + Fastify + DBOS TypeScript V3 (preview). Feels great to dogfood our new decorator-less API and explore a new web framework. Fastify's built-in validation and plugins are nice. I definitely enjoy the…
The most elegant thing about durable workflows? Observability comes for free. Durable workflows work by representing your program as database rows. There’s one row for each workflow and step, tracking its status, timing, and outcome. This info is stored for reliability, but…
I often get asked how hard it is to add durable execution to an application. Many developers get the value of durable execution. The idea is to checkpoint the state of an application to a persistent store (like a database) so that if the application ever fails it can use its…
Deduplication helps avoid redundant work that can lead to contention or wasted resources. One advantage of building queues on top of Postgres is that deduplication can naturally leverage database constraints, making the implementation both simple and efficient.
Cool new feature: queue deduplication! When you enqueue a task on a durable queue, you can set a deduplication ID. DBOS guarantees only one workflow with that ID runs at once. Where is this useful? Let’s say your workflows manage customer data, and you only want one workflow to…