unfortunately, computers
systems talks in san francisco
[ readme.txt ]
apparently “save my changes” means “make me another computer."?!
diptanu decided this was an unreasonable amount of computer. So he built copy-on-write layering into Firecracker.
we’re getting into dirty bitmaps, shared base images, and snapshotting a running Postgres image
come hang with systems people who also think saving one file should be a reasonably small request.
[ who-we-are.txt ]
unfortunately, “it works” only raises more questions.
we’re a group of systems engineers who love building products and host talks about what’s happening underneath.
we do databases, distributed systems, runtimes, networking, and infrastructure.
come share what you’re working on, ask questions, or follow a topic a few layers deeper. check out our past talks to see what we’re into.
[ past-talks.txt ]
- huffman got blamed for google's std::string::push_back by [dmitry ilyevsky (apoxy)]
- docker got the ram in the divorce by [danny lin (orbstack)]
- we ran out of h100s but the monkeys work pretty well by [sidhartha mani (constal)]
- forbidden techniques in agentic over-engineering by [haotian li]
- openai let in the webrtc trojan horse by [luke curley (moq.dev)]
- fearless until the second node by [marko vejnovic (atuin)]
- how fast can you count? by [andre bogus]
- compute with commitment issues by [andrew gazelka (indexable)]
- vibe coding the hot path by [nicolas toper (fastjvm)]
- everything failed as planned by [jordan isaacs (databricks)]
- s3 is the new disk by [almog gavra (responsive)]
- ef ef ef by [dmitriy kovalenko]