←
Daniel Worku262 posts
Daniel Worku avatar

Daniel Worku

@dbworku

Co-founder @starslingdev 💫 Previously NFLX, META and many startups

◷ Joined November 2012▧ 47 Photos and videos
Daniel Worku avatar
Daniel Worku@dbworku·21h

Cheaper CI or you get paid.

It’s time to put a stop to runaway CI bills. Introducing the CI Challenge 🤑 If @starslingdev can’t save you 10% or more on your CI bill, we’ll pay you $1,000. We’ve done over 5M CI jobs in production and have shipped hundreds of CI fixes to customer's repos, saving them over 50,000 hours in compute. Today we released new benchmarks: in 9 popular open source repos, we got an average 10x speedup in wall clock time with 70% less runner time (benchmarks linked in the post). How the challenge works: - Confirm you’re eligible + signup - Start using StarSling Runners - Start saving right away with faster runners + per second billing - Once you hit 10,000 minutes StarSling agents start sending you PRs to chop down your bill We want to help more companies remove the CI bottleneck and reduce their bill with faster runners and agents that cut down your minutes. There’s literally no reason not to do this if you’re trying to get your CI costs down. You keep whatever improvements our agents ship to you, and if we can't cut down your bill by 10% or more, you make money! How to enter the challenge:

0:19
Daniel Worku avatar
Daniel Worku@dbworku·Nov 15

aka 200 QPS 🤭

FAA acting CTO Rebecca Guy: "When we started to see the occurrence where our network was having instability, we started to look at what was the cause. We saw two places that were actually pinging our network at a very high rate -- 1,000 connections over every 4 to 5 seconds." (That sure sounds like hacking!)

1:12
Daniel Worku avatar
Daniel Worku@dbworku·Nov 13

Add one step before you use the files: - uses: starslingdev/disks@v0 with: preset: pnpm-store Good for pnpm, Cargo, Rust target folders, toolchains and massive eval datasets. Only $0.50 GB/month. What would you put on a Disk first?

Daniel Worku avatar
Daniel Worku@dbworku·Nov 13

How does it work? GitHub actions/cache stores your folder as a compressed archive. Every job needs to download that file and extract it before any file is readable. Problem: it take 26.3 seconds to access files from a 4.5 GiB folder. A Disk directly mounts the folder in 2.5 seconds (p50)

Daniel Worku avatar
Daniel Worku@dbworku·Nov 13

Introducing Disks... Persistent filesystems mounted to @starslingdev GitHub Actions jobs that are ready 10.5x faster than an actions/cache. Up to 2TB mounted per job. No practical limit to read concurrency, folder nesting depth or number of files.

0:20
Daniel Worku avatar
Daniel Worku@dbworku·Nov 3

Pro tip: for smaller DBs go no disk. Tmpfs on RAM. Much faster than even NVMe

TL;DR your database should be on local NVMe

0:45
Daniel Worku avatar
Daniel Worku@dbworku·Oct 29

Lie: "CI is hardware constrained" The top tech companies in the world are solving the great CI bottleneck of 2026 with code changes. You can too with our agent-friendly playbooks.

A new GitHub Actions runner launches every week now. But getting faster runners is just one part of what the best engineering teams do to speed up their CI. Introducing CI Playbooks ▶️ an agent-friendly library of 253 GitHub Actions optimizations from companies running software factories. We analyzed 416 pull requests and commits, and 182 blog posts, talks, and podcasts from ten companies: 1. Anthropic: CI job volume grew 25x in six months as Claude came to write 80% of its code 2. DoorDash: build and test cut from 22 minutes to 6 3. Linear: jobs 34% faster while test suites almost quadrupled 4. Mastra: 29m 56s to 5m 06s p95 on StarSling runners 5. OpenAI: about 30 CI PRs took median CI from about 10 minutes to 4.5 6. Ramp: its coding agent accounts for 74% of merged PRs 7. Shopify: an agent running experiments made the Polaris build 65% faster 8. Spotify: deployer image build went from 2h28m to about 17 minutes 9. Stripe: runs about 5% of its full Ruby test suite per build 10. Uber: CI resource usage down 53%, wait times down 37% Some of these companies aren't using GitHub Actions as their CI engine internally so we translated all the optimizations they've talked about or made publicly in their repos, into their GHA equivalents. Every technique comes with its original source and a prompt your coding agent can run on your repo. Add .md to any playbook URL for the markdown version, or just hand this URL to your agent: starsling.dev/ci-playbooks There's tons to dig into here, I'm going to highlight some of the findings I found most interesting 🧵

0:24
Daniel Worku avatar
Daniel Worku@dbworku·Oct 28

The control plane piece of Vercel CI is good. Fast booting VMs that execute each task in the optimal hardware environment concurrently.

Vercel@vercelOct 28

Introducing Vercel CI • Intelligently skips tasks with Turborepo • 2-6× faster CI in our internal repositories • Optimized for agents via Vercel MCP and CLI After nearly 8 million tasks, we're excited for the future of agentic CI. In beta on all plans. vercel.com/blog/introducing-…

Daniel Worku avatar
Daniel Worku@dbworku·Oct 23

We benchmarked lots of sandboxes and highest real-world performance products are built by teams of experts in file systems, networking, compute and hypervisors Do you know something they don’t?

Daniel Worku avatar
Daniel Worku@dbworku·Oct 23

Going bare metal is like the graveyard scene with the hyenas cackling in Lion King. You shouldn’t be there.

Daniel Worku avatar
Daniel Worku@dbworku·Oct 23

Check this out! My YC group office hours mates are killing it. Very aligned with where CI is heading.

We formally verified 123 open-source projects in a partnership with Anthropic. Formal verification at scale is here! Now we’re launching OSS Verify to scale it to the software we all depend on. We prove it’s crash-free or send you the patches it needs. Free for open-source. Let’s bootstrap the verified software ecosystem: theorem.dev/oss-verify

Yonas Beshawred avatar
⟳ Daniel Worku retweeted
Yonas Beshawred@yonasbe·Oct 22

OpenAI's CI playbook (for GitHub Actions): @OpenAIDevs cut their median CI from ~10 min to 4.5 min by asking Codex what to fix. You can do the same with your agent. We went through 855 CI workflow changes across 20 of @OpenAI's most popular repos, 120 PRs, and 33 blog posts and guides. Here's what they shipped: 1. Rank fixes by minutes saved, from your own run history 2. Fix flaky tests that fail, then pass on the same commit 3. Mock the slow dependency: a 63-test suite went from 592.6s to 29.8s 4. Run tests in fewer processes: 8m49s to 21s 5. Install prebuilt packages (plus cache fixes): 10m42s to 2m12s 6. Balance shards by measured test time: 22m26s to ~7m29s 7. Build once, test on every shard 8. Measure test concurrency: 8 jobs beat 12 9. Drop caches that cost more than they save: 66,295s of saves for 1,515s of restores 10. Put the long pole on a faster runner Each one is a copy-paste prompt for your agent: starsling.dev/blog/openai-ci