Akshay Shah
@akshayshah.org
Field CTO @ Antithesis
https://akshayshah.org
https://akshayshah.org
Baffling how the same people can be enraged by 5m builds and enthralled by multi-day async decision-making processes.
November 6, 2025 at 11:29 PM
Baffling how the same people can be enraged by 5m builds and enthralled by multi-day async decision-making processes.
Just in time for my recent re-obsession with gossip-based systems: Fly.io published a blog post on moving from multi-region Consul to Corrosion, their in-house, SWIM-based metadata store.
fly.io/blog/corrosi...
fly.io/blog/corrosi...
Corrosion
Corrosion is distributed service discovery based on Rust, SQLite, and CRDTs.
fly.io
October 28, 2025 at 12:13 AM
Just in time for my recent re-obsession with gossip-based systems: Fly.io published a blog post on moving from multi-region Consul to Corrosion, their in-house, SWIM-based metadata store.
fly.io/blog/corrosi...
fly.io/blog/corrosi...
For everything today's models are bad at, I'm really happy with how the images in this presentation came out. Starting with our designer's Halloween images and a pic of myself, then riffing from there was fun! The model feels like an in-betweener in the hand-drawn cartoon days.
October 26, 2025 at 4:58 PM
For everything today's models are bad at, I'm really happy with how the images in this presentation came out. Starting with our designer's Halloween images and a pic of myself, then riffing from there was fun! The model feels like an in-betweener in the hand-drawn cartoon days.
🧵 OK, so I just read Google's "Firefly: Scalable, Ultra-Accurate Clock Synchronization for Datacenters." Lots of cool ideas. But how do people *monitor* decentralized systems like this?
October 25, 2025 at 6:34 AM
🧵 OK, so I just read Google's "Firefly: Scalable, Ultra-Accurate Clock Synchronization for Datacenters." Lots of cool ideas. But how do people *monitor* decentralized systems like this?
Reposted by Akshay Shah
Reposted by Akshay Shah
New mascot day! Here's @akshayshah.org introducing our new mascot to a packed audience at @leaddev.com NYC. We're very sorry about what happened to the old mascot.
October 15, 2025 at 9:00 PM
New mascot day! Here's @akshayshah.org introducing our new mascot to a packed audience at @leaddev.com NYC. We're very sorry about what happened to the old mascot.
"Popularized in Haskell"
Denominators can really do a lot of work.
Denominators can really do a lot of work.
October 14, 2025 at 1:52 PM
"Popularized in Haskell"
Denominators can really do a lot of work.
Denominators can really do a lot of work.
In honor of spooky month, share a 4 word horror story that only someone in your profession would understand.
“Fifty-two bit mantissa.”
“Fifty-two bit mantissa.”
In honor of spooky month, share a 4 word horror story that only someone in your profession would understand.
"Off by one"
"Off by one"
In honour of spooky month, share a 4 word horror story that only someone in your profession would understand.
"I'm correcting your grammar"
"I'm correcting your grammar"
October 13, 2025 at 3:13 PM
In honor of spooky month, share a 4 word horror story that only someone in your profession would understand.
“Fifty-two bit mantissa.”
“Fifty-two bit mantissa.”
Datadog must have been polishing this piece of ragebait for years. It's art, really.
(But seriously, this API is like a cheese grater to the face.)
(But seriously, this API is like a cheese grater to the face.)
October 3, 2025 at 3:37 AM
Datadog must have been polishing this piece of ragebait for years. It's art, really.
(But seriously, this API is like a cheese grater to the face.)
(But seriously, this API is like a cheese grater to the face.)
Vibe-writing in Excel? No thanks 🤮
Unreliable tools can be incredibly useful - but you need a bulletproof test for their output. And spreadsheets are famously hard to get right! 94% of operational spreadsheets contain errors, and humans are exceptionally bad at catching them.
Unreliable tools can be incredibly useful - but you need a bulletproof test for their output. And spreadsheets are famously hard to get right! 94% of operational spreadsheets contain errors, and humans are exceptionally bad at catching them.
October 3, 2025 at 2:12 AM
Vibe-writing in Excel? No thanks 🤮
Unreliable tools can be incredibly useful - but you need a bulletproof test for their output. And spreadsheets are famously hard to get right! 94% of operational spreadsheets contain errors, and humans are exceptionally bad at catching them.
Unreliable tools can be incredibly useful - but you need a bulletproof test for their output. And spreadsheets are famously hard to get right! 94% of operational spreadsheets contain errors, and humans are exceptionally bad at catching them.
Best convo on the internet right now @vicvijayakumar.com
October 2, 2025 at 6:28 PM
Best convo on the internet right now @vicvijayakumar.com
Reposted by Akshay Shah
How AWS S3 serves 1 petabyte per second on top of slow HDDs // Stanislav Kozlovski
Obscure feature + obscure feature + obscure feature = bug // Michael Gibson, @antithesis.com
%CPU utilization is a lie // @brendanlong.bsky.social
Obscure feature + obscure feature + obscure feature = bug // Michael Gibson, @antithesis.com
%CPU utilization is a lie // @brendanlong.bsky.social
October 2, 2025 at 3:20 PM
How AWS S3 serves 1 petabyte per second on top of slow HDDs // Stanislav Kozlovski
Obscure feature + obscure feature + obscure feature = bug // Michael Gibson, @antithesis.com
%CPU utilization is a lie // @brendanlong.bsky.social
Obscure feature + obscure feature + obscure feature = bug // Michael Gibson, @antithesis.com
%CPU utilization is a lie // @brendanlong.bsky.social
This is *so* clever! At least in my head, this rhymes with the MapReduce philosophy of moving compute to the data.
Our F3 files embed small WASM programs to decode data. If somebody creates a new encoding and the DBMS does not have native impl, it can still read data using WASM passing Arrow buffers. Our experiments show WASM is 15-20% slower than native. We use @spiraldb.com's Vortex encoding impls.
October 1, 2025 at 5:00 PM
This is *so* clever! At least in my head, this rhymes with the MapReduce philosophy of moving compute to the data.
I’d rather write statsd to the wire by hand than descend into the Lovecraftian horror of the OTel Go docs.
September 29, 2025 at 10:37 PM
I’d rather write statsd to the wire by hand than descend into the Lovecraftian horror of the OTel Go docs.
Unexpected benefit of saving PDFs of all the papers and @oreilly.bsky.social books I’ve read over the years: I have the most awesome NotebookLM session for distributed systems.
September 28, 2025 at 10:55 PM
Unexpected benefit of saving PDFs of all the papers and @oreilly.bsky.social books I’ve read over the years: I have the most awesome NotebookLM session for distributed systems.
Reposted by Akshay Shah
Find out why AI code gen delivers only partial wins today. Join us on Oct 9 for a live webinar with @antithesis.com's Will Wilson & @akshayshah.org. They'll share how to bridge the gap between AI generation and real engineering productivity.
Don't miss it ➡️ bit.ly/Antithesis-A...
Don't miss it ➡️ bit.ly/Antithesis-A...
September 28, 2025 at 5:12 PM
Find out why AI code gen delivers only partial wins today. Join us on Oct 9 for a live webinar with @antithesis.com's Will Wilson & @akshayshah.org. They'll share how to bridge the gap between AI generation and real engineering productivity.
Don't miss it ➡️ bit.ly/Antithesis-A...
Don't miss it ➡️ bit.ly/Antithesis-A...
HBR is usually hot garbage, but you gotta respect whoever came up with “workslop” 💯
hbr.org/2025/09/ai-g...
hbr.org/2025/09/ai-g...
AI-Generated “Workslop” Is Destroying Productivity
Despite a surge in generative AI use across workplaces, most companies are seeing little measurable ROI. One possible reason is because AI tools are being used to produce “workslop”—content that appea...
hbr.org
September 28, 2025 at 5:49 AM
HBR is usually hot garbage, but you gotta respect whoever came up with “workslop” 💯
hbr.org/2025/09/ai-g...
hbr.org/2025/09/ai-g...
“Antithesis’s deterministic simulation and active exploration remove the blindfold, enabling a systematic and reproducible search for bugs.”
- Marek Siarkowicz, Kubernetes maintainer
www.cncf.io/blog/2025/09...
- Marek Siarkowicz, Kubernetes maintainer
www.cncf.io/blog/2025/09...
Autonomous Testing of etcd’s Robustness
As a critical component of many production systems, including Kubernetes, the etcd project’s first priority is reliability. Ensuring consistency and data safety requires our project contributors to…
www.cncf.io
September 25, 2025 at 4:01 PM
“Antithesis’s deterministic simulation and active exploration remove the blindfold, enabling a systematic and reproducible search for bugs.”
- Marek Siarkowicz, Kubernetes maintainer
www.cncf.io/blog/2025/09...
- Marek Siarkowicz, Kubernetes maintainer
www.cncf.io/blog/2025/09...
Shmargaret Shmitchell, indeed.
September 24, 2025 at 4:51 AM
Shmargaret Shmitchell, indeed.
Has anyone figured out how to get Claude to fix real bugs in distributed DBs? Hitting a brick wall with etcd and the long-abandoned Redis-Raft.
September 19, 2025 at 6:58 PM
Has anyone figured out how to get Claude to fix real bugs in distributed DBs? Hitting a brick wall with etcd and the long-abandoned Redis-Raft.
Excited to talk agents, AI, and enduring quality with Will and TNS.
AI gives you a coding shortcut, but the bottleneck just moves to debugging/testing.
Join @Antithesis.com CEO Will Wilson and Field CTO @akshayshah.org with TNS host @chrispirillo.bsky.social on Oct 9 to learn how to build a better workflow that integrates AI seamlessly.
➡️ bit.ly/Antithesis-A...
Join @Antithesis.com CEO Will Wilson and Field CTO @akshayshah.org with TNS host @chrispirillo.bsky.social on Oct 9 to learn how to build a better workflow that integrates AI seamlessly.
➡️ bit.ly/Antithesis-A...
September 19, 2025 at 5:31 AM
Excited to talk agents, AI, and enduring quality with Will and TNS.
So glad that testing/synctest is available without feature flags in Go 1.25!
🫡 To whichever Parse engineer originally wrote facebookgo/clock: you (and the many, many forks over the years) are finally off the hook. Thanks for paving the way.
🫡 To whichever Parse engineer originally wrote facebookgo/clock: you (and the many, many forks over the years) are finally off the hook. Thanks for paving the way.
September 15, 2025 at 12:16 AM
So glad that testing/synctest is available without feature flags in Go 1.25!
🫡 To whichever Parse engineer originally wrote facebookgo/clock: you (and the many, many forks over the years) are finally off the hook. Thanks for paving the way.
🫡 To whichever Parse engineer originally wrote facebookgo/clock: you (and the many, many forks over the years) are finally off the hook. Thanks for paving the way.
Chroma has drawn some interesting conclusions applying property-based testing to LLMs: research.trychroma.com/generative-b...
(h/t Hammad Bashir, who’s presenting this live at Chroma HQ!)
(h/t Hammad Bashir, who’s presenting this live at Chroma HQ!)
Generative Benchmarking
research.trychroma.com
September 10, 2025 at 3:03 AM
Chroma has drawn some interesting conclusions applying property-based testing to LLMs: research.trychroma.com/generative-b...
(h/t Hammad Bashir, who’s presenting this live at Chroma HQ!)
(h/t Hammad Bashir, who’s presenting this live at Chroma HQ!)
LLMs are remarkably effective at showing us - through our own automation choices! - which internal processes are just scar tissue that nobody cares about anymore.
September 6, 2025 at 5:42 AM
LLMs are remarkably effective at showing us - through our own automation choices! - which internal processes are just scar tissue that nobody cares about anymore.