Félix GV
@felixgv.ninja
Mostly here to talk about data infra. Opinions will eventually be your own.
Burry’s been wrong a few times since 2008… but is he really just a one hit wonder? Maybe this move is his big come back?
www.cnn.com/2025/11/05/b...
www.cnn.com/2025/11/05/b...
Michael Burry of ‘The Big Short’ is back with cryptic messages — and two massive bets | CNN Business
Michael Burry, the famed investor behind “The Big Short,” is betting artificial intelligence is more of a bubble than a revolution.
www.cnn.com
November 6, 2025 at 1:51 AM
Burry’s been wrong a few times since 2008… but is he really just a one hit wonder? Maybe this move is his big come back?
www.cnn.com/2025/11/05/b...
www.cnn.com/2025/11/05/b...
Very cool. And lots of new things to read up more about…
November 1, 2025 at 9:43 PM
Very cool. And lots of new things to read up more about…
🤣
The company is officially called "SO-YOU-DONT-HAVE-TO INCORPORATED'); DROP TABLE companies; --".
A lot of websites and the IRS don't like that name though.
We will announce more about it later this year. You can sign up to be on the waitlist: sydht.ai
A lot of websites and the IRS don't like that name though.
We will announce more about it later this year. You can sign up to be on the waitlist: sydht.ai
SO-YOU-DONT-HAVE-TO INCORPORATED'); DROP TABLE companies; --
SO-YOU-DONT-HAVE-TO is a next generation automated PostgreSQL optimization platform based on agentic artifical intelligence.
sydht.ai
October 23, 2025 at 4:13 PM
🤣
I often tell people about the differences of operating at various levels of the stack.
Typically, the lower you go down the stack, the slower everything goes. Development is slower, ramping to production is slower, stabilization is slower, but also decommissioning things is slower.
Typically, the lower you go down the stack, the slower everything goes. Development is slower, ramping to production is slower, stabilization is slower, but also decommissioning things is slower.
October 21, 2025 at 3:03 PM
I often tell people about the differences of operating at various levels of the stack.
Typically, the lower you go down the stack, the slower everything goes. Development is slower, ramping to production is slower, stabilization is slower, but also decommissioning things is slower.
Typically, the lower you go down the stack, the slower everything goes. Development is slower, ramping to production is slower, stabilization is slower, but also decommissioning things is slower.
How not to roll out AI: a YouTube case study
Recently (hard to say when, it was probably a gradual roll out anyway), YouTube began dubbing certain videos with a voiceover in the viewer’s language (presumably based on OS settings).
So what’s wrong? 🧵
Recently (hard to say when, it was probably a gradual roll out anyway), YouTube began dubbing certain videos with a voiceover in the viewer’s language (presumably based on OS settings).
So what’s wrong? 🧵
October 15, 2025 at 6:22 PM
How not to roll out AI: a YouTube case study
Recently (hard to say when, it was probably a gradual roll out anyway), YouTube began dubbing certain videos with a voiceover in the viewer’s language (presumably based on OS settings).
So what’s wrong? 🧵
Recently (hard to say when, it was probably a gradual roll out anyway), YouTube began dubbing certain videos with a voiceover in the viewer’s language (presumably based on OS settings).
So what’s wrong? 🧵
Pro tip: to get your PRs shipped faster, and save your teammates’ time, re-read your code, and only submit PRs in a state you wish they were in when you review them yourself 🤝
October 13, 2025 at 3:34 AM
Pro tip: to get your PRs shipped faster, and save your teammates’ time, re-read your code, and only submit PRs in a state you wish they were in when you review them yourself 🤝
Reposted by Félix GV
#VeniceDB now supports blob transfer for fast bootstrapping of instances! VeniceDB contributor Jingyan Li does a deep dive of this feature! www.youtube.com/watch?v=rosg...
Venice Blob Transfer Deep Dive
YouTube video by Venice
www.youtube.com
October 8, 2025 at 8:50 PM
#VeniceDB now supports blob transfer for fast bootstrapping of instances! VeniceDB contributor Jingyan Li does a deep dive of this feature! www.youtube.com/watch?v=rosg...
Why is it that every time I want to do something simple, I end up needing to fix some Javascript and then it’s no longer simple 🥲 ?
September 30, 2025 at 3:24 PM
Why is it that every time I want to do something simple, I end up needing to fix some Javascript and then it’s no longer simple 🥲 ?
Reposted by Félix GV
💯 This is exactly the reason we should have all the engineers on the pager. It doesn't matter if you are staff/principal. Need to be on-call.
This is very important. It is the folks who get woken up at night who are incentivized to fix things properly.
Armchair architects and village elder types may have a lot of wisdom and experience, but if they’re not *carrying the pager* then their opinion, however clever, is generally useless.
Armchair architects and village elder types may have a lot of wisdom and experience, but if they’re not *carrying the pager* then their opinion, however clever, is generally useless.
September 26, 2025 at 2:30 AM
💯 This is exactly the reason we should have all the engineers on the pager. It doesn't matter if you are staff/principal. Need to be on-call.
Trust those who carry the pager.
That’s how I summed up the work of improving reliability in a discussion today.
Conducting postmortems after incidents is a great ritual but, like any other process, it can be used either effectively or not.
That’s how I summed up the work of improving reliability in a discussion today.
Conducting postmortems after incidents is a great ritual but, like any other process, it can be used either effectively or not.
September 26, 2025 at 1:35 AM
Trust those who carry the pager.
That’s how I summed up the work of improving reliability in a discussion today.
Conducting postmortems after incidents is a great ritual but, like any other process, it can be used either effectively or not.
That’s how I summed up the work of improving reliability in a discussion today.
Conducting postmortems after incidents is a great ritual but, like any other process, it can be used either effectively or not.
Reposted by Félix GV
Good discussion on ParadeDB use cases, search patterns, and reranking.
Making transactions blend in with first-class search sounds like an interesting technical challenge… but what is the value prop?
I’m a fan of having fewer moving parts (within reason). But besides that…?
Are there cases where you actually need to, e.g., transactionally "search-modify-write"?
I’m a fan of having fewer moving parts (within reason). But besides that…?
Are there cases where you actually need to, e.g., transactionally "search-modify-write"?
September 20, 2025 at 3:41 PM
Good discussion on ParadeDB use cases, search patterns, and reranking.
Reposted by Félix GV
This is a great in-depth discussion on CDC from two actual practitioners.
I don't anticipate Brooklin being rewritten or replaced. CDC is an anti-pattern, and was always for niche problems and stopgap solutions. There's no reason to invest in it. It's been on ice for many years.
September 12, 2025 at 2:38 PM
This is a great in-depth discussion on CDC from two actual practitioners.
This looks interesting.
I know the intent is to attach it to production servers, but how about also attaching it to CI? Given a sufficiently comprehensive integration test suite, it could potentially yield much of the same benefits even pre-prod.
I know the intent is to attach it to production servers, but how about also attaching it to CI? Given a sufficiently comprehensive integration test suite, it could potentially yield much of the same benefits even pre-prod.
Runtime Library Intelligence for Production JVMs. Introducing JLib Inspector 1.0.0!
Inspect loaded JARs, unloaded JARs, their manifests, and....
Bonus: JVM details dump!
brunoborges.github.io/jlib-inspect...
Inspect loaded JARs, unloaded JARs, their manifests, and....
Bonus: JVM details dump!
brunoborges.github.io/jlib-inspect...
JLib Inspector
Runtime inventory of the libraries actually loaded by your JVM.
brunoborges.github.io
September 12, 2025 at 12:44 PM
This looks interesting.
I know the intent is to attach it to production servers, but how about also attaching it to CI? Given a sufficiently comprehensive integration test suite, it could potentially yield much of the same benefits even pre-prod.
I know the intent is to attach it to production servers, but how about also attaching it to CI? Given a sufficiently comprehensive integration test suite, it could potentially yield much of the same benefits even pre-prod.
Reposted by Félix GV
More blogging! I don't want AI agents controlling my laptop. Not because that's not useful, but because it's incredibly hard (impossible?) to do securely: sophiebits.com/2025/09/09/a...
I don’t want AI agents controlling my laptop
sophiebits.com
September 9, 2025 at 6:58 PM
More blogging! I don't want AI agents controlling my laptop. Not because that's not useful, but because it's incredibly hard (impossible?) to do securely: sophiebits.com/2025/09/09/a...
If you want to get started reading academic papers, but can’t ever seem to get through one, then let me throw you this underhand toss!
I’m sure you can finish reading this one, and then you’ll be well on your way to read more!
I’m sure you can finish reading this one, and then you’ll be well on your way to read more!
Vol:18 No:12 → Alphabets, Grammars, Calculators, and the End of Hand-Crafted Systems
👥 Authors: Idreos, Stratos
📄 PDF: https://www.vldb.org/pvldb/vol18/p5537-stratos.pdf
👥 Authors: Idreos, Stratos
📄 PDF: https://www.vldb.org/pvldb/vol18/p5537-stratos.pdf
September 8, 2025 at 8:08 PM
If you want to get started reading academic papers, but can’t ever seem to get through one, then let me throw you this underhand toss!
I’m sure you can finish reading this one, and then you’ll be well on your way to read more!
I’m sure you can finish reading this one, and then you’ll be well on your way to read more!
Reposted by Félix GV
Most software is like a banana. Postgres is pretty singular.
September 6, 2025 at 3:53 PM
Most software is like a banana. Postgres is pretty singular.
Reposted by Félix GV
Interesting article.
A bit slow-going at first, though it is a useful intro or recap for folks unfamiliar with relational data, normal form, etc.
The 3rd section is fine though somewhat obvious.
The 4th section is really where it shines. I wish more document DB users read it.
A bit slow-going at first, though it is a useful intro or recap for folks unfamiliar with relational data, normal form, etc.
The 3rd section is fine though somewhat obvious.
The 4th section is really where it shines. I wish more document DB users read it.
August 31, 2025 at 6:52 PM
Interesting article.
A bit slow-going at first, though it is a useful intro or recap for folks unfamiliar with relational data, normal form, etc.
The 3rd section is fine though somewhat obvious.
The 4th section is really where it shines. I wish more document DB users read it.
A bit slow-going at first, though it is a useful intro or recap for folks unfamiliar with relational data, normal form, etc.
The 3rd section is fine though somewhat obvious.
The 4th section is really where it shines. I wish more document DB users read it.
Been thinking about GenAI and its use cases… IMHO,
It’s appropriate for:
- Taking any skill from incompetent to mediocre.
- Getting some result without learning how to get it done.
And vice versa for what it’s not good for.
It’s appropriate for:
- Taking any skill from incompetent to mediocre.
- Getting some result without learning how to get it done.
And vice versa for what it’s not good for.
August 26, 2025 at 3:55 PM
Been thinking about GenAI and its use cases… IMHO,
It’s appropriate for:
- Taking any skill from incompetent to mediocre.
- Getting some result without learning how to get it done.
And vice versa for what it’s not good for.
It’s appropriate for:
- Taking any skill from incompetent to mediocre.
- Getting some result without learning how to get it done.
And vice versa for what it’s not good for.
Reposted by Félix GV
Hybrid Stores, WAL, Multi-WAL, Replication and CRDTs in Venice
YouTube video by Venice
youtu.be
August 23, 2025 at 3:16 AM
Reposted by Félix GV
August 20, 2025 at 2:55 PM
I'm preparing some documentation on techniques for merging batch and real-time data, where I'll explore various concepts that have emerged over the past 15 years.
August 14, 2025 at 4:19 PM
I'm preparing some documentation on techniques for merging batch and real-time data, where I'll explore various concepts that have emerged over the past 15 years.
Reposted by Félix GV
And I’m sure there are more instances out there… no way that we built this three times at LinkedIn (and without really consulting each other much, BTW) and (at least) once at Pinterest, but nowhere else.
My goal in naming it the "Hybrid Store Design Pattern" is so that we can finally talk about it.
My goal in naming it the "Hybrid Store Design Pattern" is so that we can finally talk about it.
August 14, 2025 at 2:07 PM
And I’m sure there are more instances out there… no way that we built this three times at LinkedIn (and without really consulting each other much, BTW) and (at least) once at Pinterest, but nowhere else.
My goal in naming it the "Hybrid Store Design Pattern" is so that we can finally talk about it.
My goal in naming it the "Hybrid Store Design Pattern" is so that we can finally talk about it.
Yes… it popped up in several systems over the last decade…
@venicedb.org implemented it around 2017 (some ~5 years before getting open sourced).
Other @linkedin.com proprietary systems (Search, Graph) also implemented it.
And Pinterest’s proprietary KV Store system as well.
Any others?
@venicedb.org implemented it around 2017 (some ~5 years before getting open sourced).
Other @linkedin.com proprietary systems (Search, Graph) also implemented it.
And Pinterest’s proprietary KV Store system as well.
Any others?
"Hybrid Stores are a way to implement Lambda Architectures, but by merging at write-time, rather than read-time. They can also support Kappa Architectures."
Hybrid stores are such a "duh" pattern when you see them, but it took a long time for the industry to get here.
Hybrid stores are such a "duh" pattern when you see them, but it took a long time for the industry to get here.
August 14, 2025 at 1:01 AM
Yes… it popped up in several systems over the last decade…
@venicedb.org implemented it around 2017 (some ~5 years before getting open sourced).
Other @linkedin.com proprietary systems (Search, Graph) also implemented it.
And Pinterest’s proprietary KV Store system as well.
Any others?
@venicedb.org implemented it around 2017 (some ~5 years before getting open sourced).
Other @linkedin.com proprietary systems (Search, Graph) also implemented it.
And Pinterest’s proprietary KV Store system as well.
Any others?
What is HyBRID?
HyBRID is a recursive backronym used as a mnemonic device to remember the purpose of the Hybrid Store Design Pattern. It means: Hybrid Batch/Real-time Ingested Data.
HyBRID is a recursive backronym used as a mnemonic device to remember the purpose of the Hybrid Store Design Pattern. It means: Hybrid Batch/Real-time Ingested Data.
August 12, 2025 at 8:08 PM
What is HyBRID?
HyBRID is a recursive backronym used as a mnemonic device to remember the purpose of the Hybrid Store Design Pattern. It means: Hybrid Batch/Real-time Ingested Data.
HyBRID is a recursive backronym used as a mnemonic device to remember the purpose of the Hybrid Store Design Pattern. It means: Hybrid Batch/Real-time Ingested Data.
If you’d like to learn more, this talk provides an in-depth explanation of the Hybrid Store Design Pattern:
youtu.be/hc0pgvnr3fQ
youtu.be/hc0pgvnr3fQ
June 20, 2025 at 4:34 PM
If you’d like to learn more, this talk provides an in-depth explanation of the Hybrid Store Design Pattern:
youtu.be/hc0pgvnr3fQ
youtu.be/hc0pgvnr3fQ