Raphael De Lio
@raphaeldelio.dev
Software Engineer | AI | Machine Learning | Developer Advocacy | International Speaker | Growing @ Redis (@redis.io)
https://raphaeldelio.com | https://youtube.com/raphaeldelio | https://linktr.ee/raphaeldelio
https://raphaeldelio.com | https://youtube.com/raphaeldelio | https://linktr.ee/raphaeldelio
LLMs may get expensive fast. At the same time, vector DBs (eg Redis) are versatile but underutilized
I delivered a talk at JFall yesterday about three strategies that can be implemented with vector DBs to reduce LLM calls. But the truth is that there are many more! #java
github.com/redis-develo...
I delivered a talk at JFall yesterday about three strategies that can be implemented with vector DBs to reduce LLM calls. But the truth is that there are many more! #java
github.com/redis-develo...
November 7, 2025 at 10:44 AM
LLMs may get expensive fast. At the same time, vector DBs (eg Redis) are versatile but underutilized
I delivered a talk at JFall yesterday about three strategies that can be implemented with vector DBs to reduce LLM calls. But the truth is that there are many more! #java
github.com/redis-develo...
I delivered a talk at JFall yesterday about three strategies that can be implemented with vector DBs to reduce LLM calls. But the truth is that there are many more! #java
github.com/redis-develo...
Using slidev (sli.dev) for the first time. Looking forward to delivering my first presentation with it tomorrow at J-Fall Conf in the NL - @rafaelponte.bsky.social told me about it in the beginning of the year, but took me a while to experiment with it.
November 5, 2025 at 1:44 PM
Using slidev (sli.dev) for the first time. Looking forward to delivering my first presentation with it tomorrow at J-Fall Conf in the NL - @rafaelponte.bsky.social told me about it in the beginning of the year, but took me a while to experiment with it.
Can artificial intelligence do basic math?
I was listening to Anthropic's recent video "How AI Models Think" based on their research on interpretability and found a few insights very interesting. One for example is that there's evidence that LLMs can do simple math (addition).
I was listening to Anthropic's recent video "How AI Models Think" based on their research on interpretability and found a few insights very interesting. One for example is that there's evidence that LLMs can do simple math (addition).
August 29, 2025 at 12:48 PM
Can artificial intelligence do basic math?
I was listening to Anthropic's recent video "How AI Models Think" based on their research on interpretability and found a few insights very interesting. One for example is that there's evidence that LLMs can do simple math (addition).
I was listening to Anthropic's recent video "How AI Models Think" based on their research on interpretability and found a few insights very interesting. One for example is that there's evidence that LLMs can do simple math (addition).
1/ "NOPERM User user:209 has no permissions to run the 'flushall' command"
That's the message I got when running "FLUSHALL" on my production Redis instance. 😮💨
My team and I developed an internal platform for us to track our work. The chosen database? Redis, of course.
#redis @redis.io
That's the message I got when running "FLUSHALL" on my production Redis instance. 😮💨
My team and I developed an internal platform for us to track our work. The chosen database? Redis, of course.
#redis @redis.io
August 27, 2025 at 1:54 PM
One of the things people are afraid of since the term artificial intelligence has been coined 70 years ago by John McCarthy is that it may rebel against us.
August 24, 2025 at 5:13 PM
One of the things people are afraid of since the term artificial intelligence has been coined 70 years ago by John McCarthy is that it may rebel against us.
Reposted by Raphael De Lio
I just like building stuff 🥰
August 19, 2025 at 8:50 PM
I just like building stuff 🥰
Reposted by Raphael De Lio
Let's gather to talk about #Java Gatherers.
If you're in London, please join me at the London Java Community (LJC) meet up on Monday August 18th: www.eventbrite.co.uk/e/ljc-meet-u...
If you're in London, please join me at the London Java Community (LJC) meet up on Monday August 18th: www.eventbrite.co.uk/e/ljc-meet-u...
LJC Meet-up at Redis
LJC Meet-ups is a new series of events, aimed at giving all Community members an opportunity to present at an LJC event.
www.eventbrite.co.uk
August 15, 2025 at 12:42 PM
Let's gather to talk about #Java Gatherers.
If you're in London, please join me at the London Java Community (LJC) meet up on Monday August 18th: www.eventbrite.co.uk/e/ljc-meet-u...
If you're in London, please join me at the London Java Community (LJC) meet up on Monday August 18th: www.eventbrite.co.uk/e/ljc-meet-u...
Vibe coding is the WORST IDEA of 2025 according to Dave Farley: www.youtube.com/watch?v=1A6u...
Vibe Coding Is The WORST IDEA Of 2025
YouTube video by Modern Software Engineering
www.youtube.com
August 14, 2025 at 6:58 AM
Vibe coding is the WORST IDEA of 2025 according to Dave Farley: www.youtube.com/watch?v=1A6u...
Reposted by Raphael De Lio
Since Nvidia’s primary objective is to sell more GPUs, they open-source all their research. Recently, they explored how far RL scaling can take LLMs.
ProRLv2: Expanding LLM reasoning boundaries through 3,000+ RL steps across five domains, setting a new state-of-the-art among 1.5B reasoning models.
ProRLv2: Expanding LLM reasoning boundaries through 3,000+ RL steps across five domains, setting a new state-of-the-art among 1.5B reasoning models.
August 13, 2025 at 5:15 AM
Since Nvidia’s primary objective is to sell more GPUs, they open-source all their research. Recently, they explored how far RL scaling can take LLMs.
ProRLv2: Expanding LLM reasoning boundaries through 3,000+ RL steps across five domains, setting a new state-of-the-art among 1.5B reasoning models.
ProRLv2: Expanding LLM reasoning boundaries through 3,000+ RL steps across five domains, setting a new state-of-the-art among 1.5B reasoning models.
Just because an airplane doesn’t fly the same way a bird flies, does it mean it doesn’t fly?
August 13, 2025 at 8:45 AM
Just because an airplane doesn’t fly the same way a bird flies, does it mean it doesn’t fly?
After three weeks using Claude Code, I can say vibe coding just isn’t for me.
My biggest problem is that it moves too fast. Sure, it eventually gets to the result I want, but it’s hard to follow the design choices along the way. That makes it way too easy to end up with messy, hard-to-maintain code
My biggest problem is that it moves too fast. Sure, it eventually gets to the result I want, but it’s hard to follow the design choices along the way. That makes it way too easy to end up with messy, hard-to-maintain code
August 12, 2025 at 5:38 PM
After three weeks using Claude Code, I can say vibe coding just isn’t for me.
My biggest problem is that it moves too fast. Sure, it eventually gets to the result I want, but it’s hard to follow the design choices along the way. That makes it way too easy to end up with messy, hard-to-maintain code
My biggest problem is that it moves too fast. Sure, it eventually gets to the result I want, but it’s hard to follow the design choices along the way. That makes it way too easy to end up with messy, hard-to-maintain code
Reposted by Raphael De Lio
New Redis release, but with a twist. It includes our first Rust patches to Redis Query Engine!
In particular, it ships the new TrieMap implementation I wrote with Henk Oordt, as part of @mainmatter.com's ongoing collaboration with Redis.
In particular, it ships the new TrieMap implementation I wrote with Henk Oordt, as part of @mainmatter.com's ongoing collaboration with Redis.
August 11, 2025 at 1:50 PM
New Redis release, but with a twist. It includes our first Rust patches to Redis Query Engine!
In particular, it ships the new TrieMap implementation I wrote with Henk Oordt, as part of @mainmatter.com's ongoing collaboration with Redis.
In particular, it ships the new TrieMap implementation I wrote with Henk Oordt, as part of @mainmatter.com's ongoing collaboration with Redis.
Reposted by Raphael De Lio
Using AI to analyze thousands of frames taken by drone, a mountain rescue team has found the body of a doctor who had been missing in the Alps since September 2024.
A Hiker Was Missing for Nearly a Year—Until an AI System Recognized His Helmet
Using AI to analyze thousands of frames taken by drone, The Piedmontese Alpine Rescue team has found the body of a doctor who had been missing since September 2024.
www.wired.com
August 9, 2025 at 11:56 PM
Using AI to analyze thousands of frames taken by drone, a mountain rescue team has found the body of a doctor who had been missing in the Alps since September 2024.
I see a lot of people generating images to test GPT-5 capabilities, but it’s important to note that GPT-5 does NOT generate images.
It’s still the GPT-Image-1 model released in April of this year that is generating the images.
It’s still the GPT-Image-1 model released in April of this year that is generating the images.
August 9, 2025 at 11:03 PM
I see a lot of people generating images to test GPT-5 capabilities, but it’s important to note that GPT-5 does NOT generate images.
It’s still the GPT-Image-1 model released in April of this year that is generating the images.
It’s still the GPT-Image-1 model released in April of this year that is generating the images.
Reposted by Raphael De Lio
Already better than Discover...
Great example of the unreasonable effectiveness of simple algorithms leveraging the human intelligence innate to an app with a follow graph and likes.
Conclusive proof you don't need a team of ML engineers and custom models to make a highly engaging algorithm.
Great example of the unreasonable effectiveness of simple algorithms leveraging the human intelligence innate to an app with a follow graph and likes.
Conclusive proof you don't need a team of ML engineers and custom models to make a highly engaging algorithm.
Welcome to the ✨For You✨ feed!
It finds people who liked the same posts as you, and shows you what else they've liked recently.
📌 Pin to add it to your top bar
❤️ Like the feed and repost to spread the goodness
It finds people who liked the same posts as you, and shows you what else they've liked recently.
📌 Pin to add it to your top bar
❤️ Like the feed and repost to spread the goodness
August 9, 2025 at 2:09 AM
Already better than Discover...
Great example of the unreasonable effectiveness of simple algorithms leveraging the human intelligence innate to an app with a follow graph and likes.
Conclusive proof you don't need a team of ML engineers and custom models to make a highly engaging algorithm.
Great example of the unreasonable effectiveness of simple algorithms leveraging the human intelligence innate to an app with a follow graph and likes.
Conclusive proof you don't need a team of ML engineers and custom models to make a highly engaging algorithm.
Reposted by Raphael De Lio
I'm in Brazil and had to do the same thing.
Android does not have that functionality, so I used this app.
play.google.com/store/apps/d...
I hope they add it soon.
In Brazil they do something awful, bots call you and hang up, just to check if your number is active.
Check out my block log!
Android does not have that functionality, so I used this app.
play.google.com/store/apps/d...
I hope they add it soon.
In Brazil they do something awful, bots call you and hang up, just to check if your number is active.
Check out my block log!
August 8, 2025 at 12:44 PM
I'm in Brazil and had to do the same thing.
Android does not have that functionality, so I used this app.
play.google.com/store/apps/d...
I hope they add it soon.
In Brazil they do something awful, bots call you and hang up, just to check if your number is active.
Check out my block log!
Android does not have that functionality, so I used this app.
play.google.com/store/apps/d...
I hope they add it soon.
In Brazil they do something awful, bots call you and hang up, just to check if your number is active.
Check out my block log!
Reposted by Raphael De Lio
Reposted by Raphael De Lio
Programação orientada a alfabeto
August 8, 2025 at 4:22 PM
Programação orientada a alfabeto
Reposted by Raphael De Lio
Compartilhando um pouco de conhecimento em vídeo. Recomendo 3 livros que mudaram minha carreira.
Aproveita e deixa sua recomendação livros também:
Aproveita e deixa sua recomendação livros também:
August 8, 2025 at 6:16 PM
Compartilhando um pouco de conhecimento em vídeo. Recomendo 3 livros que mudaram minha carreira.
Aproveita e deixa sua recomendação livros também:
Aproveita e deixa sua recomendação livros também:
🚀 GPT-5 is out — and that’s AMAZING news for RAG.
Every time a new model drops, people declare “RAG is dead” because of bigger context windows.
This time they’re also pointing to better accuracy with massive token counts.
Let’s talk about why they’re wrong. 🧵
Every time a new model drops, people declare “RAG is dead” because of bigger context windows.
This time they’re also pointing to better accuracy with massive token counts.
Let’s talk about why they’re wrong. 🧵
August 8, 2025 at 10:51 AM
🚀 GPT-5 is out — and that’s AMAZING news for RAG.
Every time a new model drops, people declare “RAG is dead” because of bigger context windows.
This time they’re also pointing to better accuracy with massive token counts.
Let’s talk about why they’re wrong. 🧵
Every time a new model drops, people declare “RAG is dead” because of bigger context windows.
This time they’re also pointing to better accuracy with massive token counts.
Let’s talk about why they’re wrong. 🧵
"Do you want a second opinion? We can ask Gemini."
August 7, 2025 at 5:38 PM
"Do you want a second opinion? We can ask Gemini."
"GPT-5 opens a whole new world of vibe coding."
August 7, 2025 at 5:23 PM
"GPT-5 opens a whole new world of vibe coding."
GPT 5 being announced in 10 min: www.youtube.com/watch?v=0Uu_...
Introducing GPT-5
YouTube video by OpenAI
www.youtube.com
August 7, 2025 at 4:51 PM
GPT 5 being announced in 10 min: www.youtube.com/watch?v=0Uu_...