Aditya Chadha

Code. Design. Engineering. Culture. Startups. Miami. Technology. Creativity. Director @QuriousLabs. Past: TED, Betaworks. Building:

[Aditya Chadha @ac]

"when u don't create things, u become defined by your tastes rather than ability. your tastes only narrow & exclude people. so create" _why

[Kevin Elliott @kevinelliott]

Just launched Systune (a macOS app for disk space management and performance tuning) on Better Launch.

Please help me out by upvoting on Better Launch!

See the comments.

[Solana @solana]

✅ 400ms
✅ 350ms
✅ 300ms
✅ 250ms
✅ 200ms

[Tim Devane @tdevane]

So silly, major industry self-policing never works. It's why we have police and referees.

[Aditya Chadha @ac]

@tdevane especially given the history of abuses some of these companies have 💀hard to balance innovation and regulation esp as the open weight nonUS models are miles ahead of anything homegrown

[Vipul Ved Prakash @vipulved]

Some people are interpreting this as a way for OSS models to bypass encryption of closed model APIs. If that is what you are taking away from this, please understand that the title is a massive overstatement in this context.

The point of encryption here is to keep the inference protocol stateless; I assume labs are fully aware of some side channel implications and do not care about preventing replay (why would they?) Encrypted traces offer a stateless inference protocol so inference backend can be distributed across hundreds or thousands of data centers without the need for coordinating sessions.

If your side channel is to ask the LLM "true or false" questions for the universe of words that might be contained in a trace, you'd go broke decoding a single trace. So you'd have to assume there's some super special magic trace that lived within some encrypted blob that's worth attacking in this way.

Overall definitely an interesting limitation of the encrypted vs stored traces, but you aren't going to be able to steal traces this way to train a competing model. That would cost more than building the model with the standard training arsenal.

[Pace - AI Weekly Spend Tracker @Pace_Money]

Coming to an app store near you

[Aditya Chadha @ac]

God, it’s 2026 why is personal finance software still so broken…

[Tim Devane @tdevane]

a little saturday morning lite expo writing

New Post:

Of Assholes and Obsessives

on the distorted intensities of ego and conviction

https://t.co/1bF1cpH9hp

[Aditya Chadha @ac]

@evan This is awesome. Lotta gold in them thar hills

[Evan Weaver @evan]

Put the Fauna blog's greatest hits back online: https://t.co/6dfxfGSzG7

[Tim Devane @tdevane]

i wrote this ~11 years ago, kinda feels more relevant in our agentic present

https://t.co/i4wvB9kYpb

[Aditya Chadha @ac]

@scotchi haha take that, apple!

[Aditya Chadha @ac]

Happy 2026!

[Aditya Chadha @ac]

Hmm.

[Aditya Chadha @ac]

@sbyrnes and co are legit, glad someone’s finally looking out for solo founders

[Sean Byrnes 🇺🇸 @sbyrnes]

I’m excited to finally reveal Near Horizon to the world!

In early 2023, @mfk , @mjacobstein and I were frustrated at how great solo founders were told to “go find a co-founder” by traditional VC firms ...

...so set out to create a new kind of early stage venture firm!

1/

[Sean Byrnes 🇺🇸 @sbyrnes]

I’m excited to finally reveal Near Horizon to the world!

In early 2023, @mfk , @mjacobstein and I were frustrated at how great solo founders were told to “go find a co-founder” by traditional VC firms ...

...so set out to create a new kind of early stage venture firm!

1/

[Aditya Chadha @ac]

New season of the bear is 🔥 so far

[Aditya Chadha @ac]

@sgrove what do you like most about it over ts?

[Aditya Chadha @ac]

Holy fuck. This is epic.

[Vipul Ved Prakash @vipulved]

OpenAI API compatibility shipped for 100+ models on @togethercompute API.

Replace GPT calls with Mixtral or Llama-70B, get faster responses and for less $$

🚀🚀🚀

[Aditya Chadha @ac]

@Cleartrip indigo made me pay for seats again, could you refund me for seat selection? Trip Id: 230922126144