Praveen

nomad/aviation nerd | shipping agents @getpostman @avgeekinc | ✈️ | 🌐 | 🎨 | never settle at the foothills.

[Bangalore Aviation @BLRAviation]

In late 2025, an uncommanded pitch-down on a JetBlue A320 led to swift a global emergency AD over solar flare-induced software corruption.

The world’s airlines grounded fleets and swiftly updated their software.

Contrast that with the #AAIB preliminary report on Air India VT-EXO (AI2379) from Phuket to Delhi. The aircraft suffered a sudden loss of all three hydraulic systems (Green, Blue, Yellow) resulting in injured passengers and crew.

Yet a month later, there is no visible urgency on this potentially catastrophic technical failure.

Instead, the report’s sole interim recommendation targets psychoactive substance testing after one of the pilots tested non-negative.

Why is this technical issue not commanding the world’s airlines’ attention?

https://t.co/tpCTsEwqlb

[Praveen @praveentcom]

NFC payments are working since very long in India and in INR currency. I've been using my Brex and Mercury cards in all POS machines since last 5 years.

Merchants who has INR Stripe account were also able to use Apple Pay regardless since years, but nice to see it's getting into other PGs as well now.

https://t.co/dT1e9Als6f

[Praveen @praveentcom]

My first Apple Pay NFC transaction. Probably the first in Coimbatore too.

Ice cream time 😋🍨🇮🇳

@brexHQ seamless!

[Praveen @praveentcom]

@ani_digital @ANI "Zero tolerance" and "100% voluntary" doesn't go together, what am I missing?

[Praveen @praveentcom]

@Fahadnaimb @BrightFuts Is there a possibility of any structural damage apart from the nose? I can see it uprooted the pole, so assuming the force was good enough.

[Praveen @praveentcom]

@thsottiaux Yes, need a reset.

[Praveen @praveentcom]

@AbiramiSukumara @Google Congratulations @AbiramiSukumara! 🎉

[Praveen @praveentcom]

Learning curve was the blocker to use React Native so far for me. With AI, now I am moving my apps from SwiftUI/Compose to RN.

A shared Rust engine for the models & service layer and RN for the presentation layer is doing great.

[Praveen @praveentcom]

Had a minor incident with my Slavia where the ASC recommended replacing a headlight. I did with insurance and it costed me 4k in insurance charges. 20k is high - which insurance is this?

Ensure there aren’t any hairline cracks on the headlights. It leads up to fog build overtime which limits the light intensity harshly at certain conditions.

[Praveen @praveentcom]

TIL. If you've connected the phone via iPhone Mirroring, live activities show up in the menu bar...

[Praveen @praveentcom]

@AP Impressive but looks hilarious! 😅

[Praveen @praveentcom]

@juhi_singh15 @ayushshekhar17 You should try Muir Woods National Park. You’ll feel so refreshed and relaxed.

Pro tip: You won’t get cellphone reception, make sure to book to/fro cab.

https://t.co/ZgL70la1ms

[Praveen @praveentcom]

Thank you, Codex.

[Praveen @praveentcom]

@gregpr07 They are aware of this and want it this way. I asked about this four months back, this was the response.

[Celso Martinho @celso]

@PraveenTcom @Cloudflare This is good, not bad. Browser Run isn’t trying to hide its identity or circumvent website policies. Quite the opposite, actually. https://t.co/MJ50E0bAHM and https://t.co/S1ZMX1Goac

[Praveen @praveentcom]

@nthglsn @gregpr07 Which is insanely costly and beats the purpose of these tools.

[Praveen @praveentcom]

@Vgdktk Spill the tea please

[Praveen @praveentcom]

@msomuin I found it insanely slow for some reason. Not a model issue for sure, harness seems complicated.

[Praveen @praveentcom]

@danielcraft @migtissera I don’t think the source code should be considered telemetry in any way.

[Praveen @praveentcom]

cc @_gongy

[Praveen @praveentcom]

@modal hey team, I created an endpoint for Kimi K3 (managed) and tried setting it up for my workflow.

I got quite a few 429s immediately and is occurring frequently. Not even a total of 100 requests yet - can we take a look?

[Kimi.ai @Kimi_Moonshot]

Releasing the model weights and technical report of Kimi K3.

Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window.

New model architecture: 2.5x the intelligence per unit of compute, not just more params.

Alongside Kimi K3, we're opening up more of the stack behind it — high-performance attention kernels, MoE communication library, and infrastructure for running agent environments at scale.

Model weights: https://t.co/7m7eEg6Y0B
Tech report: https://t.co/yeu6cjpMCT
Tech blog: https://t.co/YTfiMSNM1f