Technically retweeted
My timeline is full of RL. RL environments, RL gyms, data labeling companies. Startups scaling to $1B in revenue in months. What is going on?
Post-training seems like the new frontier. Reinforcement learning is the emerging preferred method, with an ecosystem worth $Bs growing s it work? And how much does it really add to good pre-training?
On Ep 3 of Technically Speaking from @readtechnically, a rare conversation from both sides of the RL stack:
@slippylolo (co-founder & CEO of Adaptive ML, now at @datadoghq ) and @nerdsane (Datadog):
→ Why a monitoring company bought an RL startup: "At its heart, this is an observability problem."
→ Whether your startup should train its own model: "for 99% of people it doesn't make any sense to do that"
→ Creating synthetic training data: "You cannot create data ex nihilo and hope for the best."
→ What actually makes for great post-training: "it's mostly a question of high quality execution"
Post-training seems like the new frontier. Reinforcement learning is the emerging preferred method, with an ecosystem worth $Bs growing s it work? And how much does it really add to good pre-training?
On Ep 3 of Technically Speaking from @readtechnically, a rare conversation from both sides of the RL stack:
@slippylolo (co-founder & CEO of Adaptive ML, now at @datadoghq ) and @nerdsane (Datadog):
→ Why a monitoring company bought an RL startup: "At its heart, this is an observability problem."
→ Whether your startup should train its own model: "for 99% of people it doesn't make any sense to do that"
→ Creating synthetic training data: "You cannot create data ex nihilo and hope for the best."
→ What actually makes for great post-training: "it's mostly a question of high quality execution"
8
4
31
1,981
Just ship it, benchmarks be damned:
"You need to ship the best of what you have at any given time."
"You need to ship the best of what you have at any given time."
0
0
3
133
Show me the numbers!
"If I don't have numbers, I don't believe you. Like, why should I believe you?"
"If I don't have numbers, I don't believe you. Like, why should I believe you?"
1
0
2
188
"You never quite know what you're going to get when you push the frontier."
But what does pushing the frontier actually look like day-to-day? What do AI researchers...actually do?
On Ep 2 of Technically Speaking, @jefrankle (Chief AI Scientist @databricks) & @meganreyno get intoteam: "If I don't have numbers, I don't believe you."
→ Why you should ship before you feel ready: "you need to ship the best of what you have at any given time"
→ Who he hires: "people who are never quite satisfied"
🎬 Watch the full episode: youtu.be/Qz58c13N7ms
Presented by our friends at @Railway.
Stay tuned for some of our favorite moments...
But what does pushing the frontier actually look like day-to-day? What do AI researchers...actually do?
On Ep 2 of Technically Speaking, @jefrankle (Chief AI Scientist @databricks) & @meganreyno get intoteam: "If I don't have numbers, I don't believe you."
→ Why you should ship before you feel ready: "you need to ship the best of what you have at any given time"
→ Who he hires: "people who are never quite satisfied"
🎬 Watch the full episode: youtu.be/Qz58c13N7ms
Presented by our friends at @Railway.
Stay tuned for some of our favorite moments...
3
1
11
4,546
"I would start this exact company today, right now, to do the same thing for agents. The opportunity for us and the TAM has gotten larger and larger. I think that's mostly luck. I did not anticipate that, to be clear."
- @grinich, Founder & CEO at @WorkOS, on Technically Speaisode: youtu.be/2I8pJt9kXa8
- @grinich, Founder & CEO at @WorkOS, on Technically Speaisode: youtu.be/2I8pJt9kXa8
0
2
3
454
"If you start building a product and everybody is building the same thing and it's like consensus, there's no way you can survive. You need to be kind of misunderstood for a while."
- @grinich on ep 1 of Technically Speaking
📌 Watch the full episode: youtu.be/2I8pJt9kXa8
- @grinich on ep 1 of Technically Speaking
📌 Watch the full episode: youtu.be/2I8pJt9kXa8
0
1
2
256
"How many people do you need on your team?"
Michael Grinich (@grinich), Founder & CEO at @WorkOS, on how they hired a head of finance:
"'I have, like, eight or nine people.' And then they'd kind of say under their breath, 'But I think I can do it with just two.' It just means we wor in an office building full of people just doing calculations, like we used to for bookkeeping."
Watch the full episode: youtu.be/2I8pJt9kXa8
Michael Grinich (@grinich), Founder & CEO at @WorkOS, on how they hired a head of finance:
"'I have, like, eight or nine people.' And then they'd kind of say under their breath, 'But I think I can do it with just two.' It just means we wor in an office building full of people just doing calculations, like we used to for bookkeeping."
Watch the full episode: youtu.be/2I8pJt9kXa8
0
0
4
345
"In practice, it needs access to everything. But also nothing."
We thought access control was a solved problem, then agents showed up.
On ep 1 of Technically Speaking, @grinich (founder of @WorkOS) & @meganreyno discuss:
→ Why AI agents require rethinking Auth: "It's kind of likeyour organization very quickly."
→ What's stopping companies from letting agents cook (it's not the quality of models or harnesses): "To err is human, but to screw up 10,000 times per second you need a computer."
→ How Michael is hiring differently bc of AI
🎥 Watch the full episode: youtu.be/2I8pJt9kXa8
Presented by our friends at @Railway.
Original music by The Jazz Boys (don't sleep on them).
We thought access control was a solved problem, then agents showed up.
On ep 1 of Technically Speaking, @grinich (founder of @WorkOS) & @meganreyno discuss:
→ Why AI agents require rethinking Auth: "It's kind of likeyour organization very quickly."
→ What's stopping companies from letting agents cook (it's not the quality of models or harnesses): "To err is human, but to screw up 10,000 times per second you need a computer."
→ How Michael is hiring differently bc of AI
🎥 Watch the full episode: youtu.be/2I8pJt9kXa8
Presented by our friends at @Railway.
Original music by The Jazz Boys (don't sleep on them).
3
5
15
1,177
it seems like every week the AI wave exposes a new set of technical terms that we should know about.
so we phoned a friend (@meganreyno), and this Thursday we’re launching Technically Speaking - bringing in technical founders & leaders to break down the technical problem theyng over.
the Season 1 lineup is downright effervescent:
- Michael Grinich (WorkOS) on Auth
- Jonathan Frankle (Databricks) on AI research
- Julien Launay and Sesh Nalla (Datadog) on RL
- Jake Cooper (Railway) on Cloud
- and five (5!) more special guests to be announced
new episodes posting every Thursday until the holidays.
who should we have on next?
so we phoned a friend (@meganreyno), and this Thursday we’re launching Technically Speaking - bringing in technical founders & leaders to break down the technical problem theyng over.
the Season 1 lineup is downright effervescent:
- Michael Grinich (WorkOS) on Auth
- Jonathan Frankle (Databricks) on AI research
- Julien Launay and Sesh Nalla (Datadog) on RL
- Jake Cooper (Railway) on Cloud
- and five (5!) more special guests to be announced
new episodes posting every Thursday until the holidays.
who should we have on next?
12
7
35
7,097
ai writes code at light speed now, so github invented a way for humans to pretend they're keeping up
0
0
0
263
Databases aren't glamorous, but they're the most likely piece of the stack to ruin the launch of your vibe-coded app.
- A database is like a spreadsheet, except with more rules on how to interact with it.
- The schema is the rulebook: what every column can hold, and how other.
- Changing it on a live app is a migration, which must be done with care.
- Big files don't go in the database, they go into object storage.
Learn more about the care and feeding of databases in pt 2 of our series "Software Engineering for Vibe Coders": technically.dev/posts/database…
- A database is like a spreadsheet, except with more rules on how to interact with it.
- The schema is the rulebook: what every column can hold, and how other.
- Changing it on a live app is a migration, which must be done with care.
- Big files don't go in the database, they go into object storage.
Learn more about the care and feeding of databases in pt 2 of our series "Software Engineering for Vibe Coders": technically.dev/posts/database…
0
0
4
325
your AI won't tell you this stuff but we will
follow @readtechnically for more 👾
follow @readtechnically for more 👾
0
0
0
121
your AI won't tell you this stuff but we will
follow @readtechnically for more 👾
follow @readtechnically for more 👾
0
0
1
163