Hauntylicious

America runs on ffmpeg

------
tweets are my own and do not reflect the opinions of my employer

[Hauntylicious @hauntylicious]

How a 10mph wind feels like on the beach vs on the golf course.

[Hauntylicious @hauntylicious]

@GergelyOrosz My review style has changed to attempt to compensate. Volume is a significant stressor. E2E tests are king. Where AI goes wrong now is not getting clear rules from devs or adding backups for fear of failing. AI will have your system 200 OK and silently up in flames doing nothing.

[Roger Steele @RogerSteeleJr]

Tournament Golf is King. If you love golf go and sign up for one. It’ll change your life.

[Hauntylicious @hauntylicious]

@siddharthkp Just got here. Going to enjoy the view for awhile 🫡

[Hauntylicious @hauntylicious]

@DoerflerJared @HannaGolfCo Graphite 👀

[Hauntylicious @hauntylicious]

All that is to say, open source is more important than ever. Given what happened to tailwind though, the path forward isn't as clear. It will be interesting to see the evolution of on prem or privately hosted models to deal with the rug pulling

[Hauntylicious @hauntylicious]

Nature gonna nature. The incentive for private and public companies at its core is internal. It is what it is and no one should be surprised here. If you aren't actively regression testing AI workloads on foundational models regularly you're going to get caught in the wake.

[alphaXiv @askalphaxiv]

As believers of open research, we are disappointed to see Anthropic silently degrading Fable 5 for AI development

"Any topic related to building pretraining pipelines, distributed training infrastructure, or ML accelerator design... may have limited effectiveness through Claude via methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning."

Not only do they get to decide what you use LLMs for in research, but this also enables them to silently intervene in your research without you knowing.

This sets a dangerous precedent. If a model refuses openly, users can understand the boundary. If a model falls back to another model, users can still evaluate the difference. But if a model silently modifies or weakens its own answers while still pretending to help, researchers lose the ability to know whether a failed result came from their own idea, their implementation, or an invisible intervention by the model provider.

That is not safety. Safety policies should be transparent, auditable, and user-visible.

On top of that, the people most harmed by this are not the largest labs with massive teams and proprietary infrastructure. It is the independent researchers, academic groups, startups, and open-source builders who rely on public tools to compete, innovate, and pioneer AI for everyone else.

[Hauntylicious @hauntylicious]

@DoerflerJared @HannaGolfCo Got one 👌👌

[Jared Doerfler @DoerflerJared]

Let's give this putter away.

Here's how to win.

• Retweet this tweet
• Fill out the form in the next tweet

Winner can choose the below.

• 3 lie angles (69,70,71)
• 5 sight line options
• RH or LH
• Length
• Chrome or Black shaft

Runs through 5/25 at 5:00 Central. Winner will be selected on 5/25 by 10:00 Central.

[Hauntylicious @hauntylicious]

@Mike_kim714 @Titleist Winner gets a swing roast 👀? Sign me up.

[Michael S. Kim @Mike_kim714]

500k followers giveaway pt 1!

My golf bag plus some @Titleist goodies

Comment, like, repost to enter. Must be a follower

Clubs not included unfortunately*
Still need those for my day job

[Hauntylicious @hauntylicious]

I volunteer as tribute. I play some club events and attempt to qualify for things here and there but haven't played in anything noteworthy.

I would genuinely be concerned for patron safety on the first tee. Any score in the first 9 holes under double digits would be a success.

[Arnie McNair 🇺🇸 @therealmcnair]

Here was the debate last night: non tournament tested club golfer with low single digit GHIN starts the masters at -100. Does he win?

My answer is a strong no chance.

[Hauntylicious @hauntylicious]

Bigger bailout than the 08 financial crisis in that #badgers basketball game

[Hauntylicious @hauntylicious]

I'm not genius but maybe cover Loveland before worrying about anything else.

[Hauntylicious @hauntylicious]

Lolololol

[Hauntylicious @hauntylicious]

Same game different day. Yawn. Win or lose from here doesn't matter. Lafleur is getting extended and I can't think of any reason why we won't be seeing this seesaw offense for years to come. I hope I'm wrong.

[Hauntylicious @hauntylicious]

Whats old is new again (positive!)

Building redundant features / costs / optimizations due to simply not knowing what other people are doing.

It used to be just code duplication... now its prompts and agents.

Build in public. Talk. Share.

Same job, different medium.

[Benjamin Verbeek @benjaminvrbk]

I'm an engineer at Lovable and I spent the holidays improving our system prompt.

Lovable is now 4% faster and a lot better at creating good designs. The crazy part is that this also ended up decreasing our LLM costs by $20M per year!

Below is how I did it.

[Benjamin Verbeek @benjaminvrbk]

I'm an engineer at Lovable and I spent the holidays improving our system prompt.

Lovable is now 4% faster and a lot better at creating good designs. The crazy part is that this also ended up decreasing our LLM costs by $20M per year!

Below is how I did it.

[Hauntylicious @hauntylicious]

@packers #ProBowlVote + Jordan Love

[Green Bay Packers @packers]

🚨 Signed Jordan Love jersey giveaway 🚨

To win:
1. RT
2. Comment #ProBowlVote + Jordan Love

[Hauntylicious @hauntylicious]

I'm not a betting man but, based off my 5 days of usage, the entirety of SF YC is being built by Gemini 3 Pro because around 9-10am PT it starts grinding to a hault 😉

[Cursor @cursor_ai]

Gemini 3 Pro is now available in Cursor!