heyhey @_janbjj

AI Engineer | bjj brown belt 🥋 | Joined May 2009

Tweets

3K
Followers

418
Following

1K
Likes

6K

antirez @antirez

3 days ago

I see very worrying US companies executives declarations depicting open weight models as a risk. How much sold to the most furious capitalistic vision you need to be, to really believe it is better that a few companies control AI for all the world? They said the same for OSS.

62 106 1K 40K 54

View Details

heyhey @_janbjj

2 days ago

@zeroxkyle training data poisoning, state govt could rewrite history

0 0 0 23 0

View Details

heyhey @_janbjj

3 days ago

@cerebras great insights, thanks for sharing this

0 0 1 476 0

View Details

Kimi.ai @Kimi_Moonshot

4 days ago

Meet Kimi K3

743 3K 31K 10.0M 7K

View Details

heyhey @_janbjj

2 weeks ago

😂

Yuchen Jin @Yuchenj_UW

2 weeks ago

Didn’t know Zuck had a Twitter account. His last tweet was in 2023. Pretty much confirms X is the best place to talk about AI.

167 248 7K 640K 504

0 0 1 163 0

View Details

heyhey @_janbjj

2 weeks ago

@businessbarista AI For use case identification i have a project live that has a dataset of ~7000 job postings growing daily that is searchable. has graph cohorts of tools used etc Features im currently working on getting my agents do deep data analytics work with pandas etc.. and reports back

0 0 0 37 0

View Details

Mitchell Hashimoto @mitchellh

3 weeks ago

Amongst my friends, Spotify is the lowest quality consumer app we still pay for. It certainly hasnt gotten noticeably better in the last couple years (arguably worse). So, this is not the positive look Ant and Spotify are spinning here. Bigger picture, this is the problem with a lot of AI reporting. It reports completely meaningless metrics like deploys per day or LoC. Why don’t we start reporting consumer satisfaction reports? Actually end state research results. All the no nuance AI people always come out and think that this is anti AI. Again, I think AI is great and Claude is great. But this is bad marketing and makes both look like clowns.

ClaudeDevs @ClaudeDevs

3 weeks ago

Boris sat down with Spotify VP of Engineering Niklas Gustavsson. Spotify ships 4,500 production deploys a day, and 73% of PRs are now AI-assisted.

539 233 4K 5.5M 3K

349 573 8K 665K 980

View Details

heyhey @_janbjj

3 weeks ago

@theo the good thing about this is that you can onboard a new dev in minutes with pre configured containers and all AIOps set up already

0 0 0 6 0

View Details

heyhey @_janbjj

3 weeks ago

@jietang improve token efficiancy and make it faster

0 0 0 9 0

View Details

Svag @Svaghost

3 weeks ago

@banteg Watching the thinking blocks of GLM 5.2 feels like being inside the head of a schizophrenic... Thin line between genius and lunatic, I suppose

1 1 2 261 0

View Details

Yuchen Jin @Yuchenj_UW

3 weeks ago

DeepSeek is the GOAT. 🐳 They just published DSpark, a new speculative decoding method that boosts throughput by 51% to 400%. They also open-sourced DeepSpec, the training framework behind it. This is the real open AI.

104 456 4K 370K 1K

View Details

geoff @GeoffreyHuntley

4 weeks ago

Some personal hot takes from AI: engineer Miami follows... 1. Software development is a dead-end profession because anyone can be a software developer now. 2. Anyone can use Cursor or any other tool and generate code. Being a coder and being a software engineer are different. 3. Computers used to be gated; now everyone has the power to make computers malleable. Everyone is a software developer now, but that does not mean they are software engineers 4. If you cannot demonstrate how a coding agent works, you are just a consumer and have imposed an artificial glass ceiling on your career as a software engineer. 5. If you are curious, you will have a job. If you have not been curious in the last two years, you are replaceable. 6. SaaS per-seat economics may become unstable as customers need fewer people to achieve results, prompting founders to think about new unit economics 7. Most companies will take two or three years (or more!) to figure out AI transformation. 8. Some companies are already building AI native teams of five to ten people who can build with the grain of AI 9. There will be an explosion in the number of software developers. Software development is now essentially free, and tokens are cheaper than humans 10. Not enough engineers know what it means to be a product engineer 11. JIRA ticket monkeys are cooked 12. If your company has banned AI, you should quit that company 13. AI is more like a musical instrument than just a tool play with it, make discoveries, build intuition learn where AI is good and where it fails

54 57 391 58K 336

View Details

heyhey @_janbjj

4 weeks ago

@VittoStack interested, working on a test suite currently

1 0 1 42 0

View Details

heyhey @_janbjj

4 weeks ago

@antirez unfortunately thats 2 years away

0 0 0 406 0

View Details

heyhey @_janbjj

4 weeks ago

@alisawuffles thank you for sharing

0 0 0 20 0

View Details

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex

a month ago

I have tried to use OpenRouter Fusion API with cheap open models only, and saw reasoning that surpasses any of them individually. Then I looked into API logs and saw that this "Fusion" still calls Opus 4.8 as a judge. I see no way to disable it. Not cool, OpenRouter. Not cool.

OpenRouter @OpenRouter

a month ago

Introducing the Fusion API, the smartest compound model in the market. Fusion achieves Fable-level intelligence at half the price. How it works 👇

725 2K 15K 6.3M 13K

50 33 1K 132K 321

View Details

heyhey @_janbjj

a month ago

impressive

OpenRouter @OpenRouter

a month ago

Introducing the Fusion API, the smartest compound model in the market. Fusion achieves Fable-level intelligence at half the price. How it works 👇

725 2K 15K 6.3M 13K

0 0 1 19 0

View Details

heyhey @_janbjj

a month ago

@DonAlt retail buying tops stocks are mirroring past crypto cycles 1:1

0 0 0 8 0

View Details

Sayash Kapoor @sayashk

a month ago

There is a lot of justified anger at Anthropic for sandbagging Fable 5 for AI development tasks. But an unanticipated side effect is that third-party evaluators can no longer credibly use the model for evaluations. Case in point: we are in the middle of running *really hard* AI R&D evaluations. Fable 5 would be a perfect test candidate. But because of Anthropic's guardrails, we can't know if the model failed or if their classifiers blocked the capability. By the way, this is not just true for AI R&D. Since Anthropic doesn't make it clear when they are sandbagging, this could seep into any number of technical tasks, and the evaluators wouldn't have any way to know. So they can't credibly claim to evaluate state-of-the-art accuracy using the model.