Arnav Jaitly

4.6K posts

Arnav Jaitly banner
Arnav Jaitly

Arnav Jaitly

@arnitly

NLP | LLM | Trying to make sense of AI before it outsmarts me

London, England Katılım Aralık 2017
567 Takip Edilen613 Takipçiler
Sabitlenmiş Tweet
Arnav Jaitly
Arnav Jaitly@arnitly·
One of the most compelling ideas I encountered this year was the ancient Greek perception of time in two distinct forms: ‘Kronos’ and ‘Kairos’. Their contrast helped me understand a feeling I’ve carried for years but never quite named: time anxiety, the persistent fear that I’m not making the most of my time. In English, we have only one word for time, but the Greeks were more precise. Kronos refers to chronological, measurable time. Think of it as the hours, minutes, and seconds that structure our schedules and calendars. It is the version of time that moves in a straight line, divided into boxes we often feel pressured to fill with as many tasks, milestones, and achievements as possible. Kairos, on the other hand, is almost its opposite. It describes the quality of time rather than its quantity: the elastic, subjective experience of moments that sometimes rush by and sometimes stretch out endlessly. Kairos invites depth instead of speed. When we inhabit Kairos, we stop tallying productivity and instead ask ourselves whether we feel present, engaged, and alive. And in doing so, time anxiety loosens its grip. The distinction matters because we live in a profoundly Kronos-driven society. Our calendars look like pieces of lego. Infact, there is an unspoken Kronos timeline for our lives too (when we’re expected to buy a house, get married, have children and so on). Falling “off-schedule” often carries a sense of shame, as if failing to fill the boxes on time means failing overall. Kronos makes us efficient, but it also make us robotic. I realized that our mental well-being tends to flourish in Kairos. Organic, immersive experiences such as creativity, connection, presence don’t obey clocks. They can’t be forced into evenly sized blocks. And while it might be impossible to live in Kairos constantly, the challenge is not to reject Kronos but to carve out moments of Kairos within it. The easiest way I have implemented this in my life is asking myself this question - “What is a non-quantifiable way of measuring success today?” My answers to this question range from learning a new song on guitar, going for a long walk without any gadget, cooking a new meal, or just yapping with my friends for 2 hours. As a result, I feel the best way to optimize my time is to find ways to build Kairos into my life deliberately: choosing depth over speed, quality over quantity, and aliveness over mere productivity. Living between Kronos and Kairos is not about choosing one over the other but about learning to move intentionally between them, measuring our days not just by what we accomplish, but by how we live them.
English
0
0
2
192
Arnav Jaitly retweetledi
Hemant Mohapatra
Hemant Mohapatra@MohapatraHemant·
Just blows my 🤯 that in the last few weeks @Airbound_Aero has delivered 700+ samples for @NarayanaHealth, one of India's largest hospital networks. ETA down from 1.5hrs to <4mins. Zero crashes. Revolutionary. The asymptote to 1M+ deliveries has started.
English
35
192
1.7K
82.3K
Arnav Jaitly retweetledi
emergence.ai
emergence.ai@emergence_ai·
Today, we’re excited to introduce Emergence India Labs: India’s first dedicated frontier AI lab focused on autonomous agents, based in Bengaluru.
emergence.ai tweet media
English
8
19
149
41.6K
Arnav Jaitly retweetledi
Sundar Pichai
Sundar Pichai@sundarpichai·
We trained a new flood forecasting model designed to predict flash floods in urban areas up to 24 hours in advance. To help address a flash floods data gap, we created Groundsource: a new AI methodology using Gemini to identify 2.6M+ historical events across 150+ countries. We’re open-sourcing this dataset to advance global research, and urban flash flood forecasts are live now in Flood Hub to help communities stay safe.
Sundar Pichai tweet media
English
304
1.4K
11.3K
806.6K
Arnav Jaitly retweetledi
Todd Saunders
Todd Saunders@toddsaunders·
The token cost to build a production feature is now lower than the meeting cost to discuss building that feature. Let me rephrase. It is literally cheaper to build the thing and see if it works than to have a 30 minute planning meeting about whether you should build it. It’s wild when you think about it. This completely inverts how you should run a software organization. The planning layer becomes the bottleneck because the building layer is essentially free. The cost of code has dropped to essentially 0. The rational response is to eliminate planning for anything that can be tested empirically. Don’t debate whether a feature will work. Just build it in 2 hours, measure it with a group of customers, and then decide to kill or keep it. I saw a startup operating this way and their build velocity is up 20x. Decision quality is up because every decision is informed by a real prototype, not a slide deck and an expensive meeting. We went from “move fast and break things” to “move fast and build everything.” The planning industrial complex is dead. Thank god.
English
374
569
5.5K
462.8K
Arnav Jaitly
Arnav Jaitly@arnitly·
Kind of insane that @ManusAI doesn’t have an option to log in via @Meta despite getting acquired by them.
Arnav Jaitly tweet media
English
0
0
0
33
Jelam
Jelam@jelambhatt·
the vibe really is shifting. in the spirit of londonmaxxing, excited to announce that we’re resuming @londontechcoll meet-ups in april 🇬🇧 join our next edition to meet diverse tech folks that are long london too 🚀 follow @londontechcoll or leave a comment to receive updates 👇
English
9
2
51
3.5K
Arnav Jaitly
Arnav Jaitly@arnitly·
I don’t care if we win or lose, that was a monster effort by the team. What a one sided innings! Total dominance by the men in blue!
English
0
0
1
38
Arnav Jaitly retweetledi
Sebastian Raschka
Sebastian Raschka@rasbt·
While waiting for DeepSeek V4 we got two very strong open-weight LLMs from India yesterday. There are two size flavors, Sarvam 30B and Sarvam 105B model (both reasoning models). Interestingly, the smaller 30B model uses “classic” Grouped Query Attention (GQA), whereas the larger 105B variant switched to DeepSeek-style Multi-Head Latent Attention (MLA). As I wrote about in my analyses before, both are popular attention variants to reduce KV cache size (the longer the context, the more you save compared to regular attention). MLA is more complicated to implement, but it can give you better modeling performance if we go by the ablation studies in the 2024 DeepSeek V2 paper (as far as I know, this is still the most recent apples-to-apples comparison). Speaking of modeling performance, the 105B model is on par with LLMs of similar size: gpt-oss 120B and Qwen3-Next (80B). Sarvam is better on some tasks and worse on others, but roughly the same on average. It’s not the strongest coder in SWE-Bench Verified terms, but it is surprisingly good at agentic reasoning and task completion (Tau2). It’s even better than Deepseek R1 0528. Considering the smaller Sarvam 30B, the perhaps most comparable model to the 30B model is Nemotron 3 Nano 30B, which is slightly ahead in coding per SWE-Bench Verified and agentic reasoning (Tau2) but slightly worse in some other aspects (Live Code Bench v6, BrowseComp). Unfortunately, Qwen3-30B-A3B is missing in the benchmarks, which is, as far as I know, is the most popular model of that size class. Interestingly, though, the Sarvam team compared their 30B model to Qwen3-30B-A3B on a computational performance analysis, where they found that Sarvam gets 20-40% more tokens/sec throughput compared to Qwen3 due to code and kernel optimizations. Anyways, one thing that is not captured by the benchmarks above is Sarvam’s good performance on Indian languages. According to a judge model, the Sarvam team found that their model is preferred 90% of the time compared to others when it comes to Indian texts. (Since they built and trained the tokenizer from scratch as well, Sarvam also comes with a 4 times higher token efficiency on Indian languages.
Sebastian Raschka tweet media
Pratyush Kumar@pratykumar

📢 Open-sourcing the Sarvam 30B and 105B models! Trained from scratch with all data, model research and inference optimisation done in-house, these models punch above their weight in most global benchmarks plus excel in Indian languages. Get the weights at Hugging Face and AIKosh. Thanks to the good folks at SGLang for day 0 support, vLLM support coming soon. Links, benchmark scores, examples, and more in our blog - sarvam.ai/blogs/sarvam-3…

English
45
706
4.1K
251K
Arnav Jaitly retweetledi
Pratyush Kumar
Pratyush Kumar@pratykumar·
📢 Open-sourcing the Sarvam 30B and 105B models! Trained from scratch with all data, model research and inference optimisation done in-house, these models punch above their weight in most global benchmarks plus excel in Indian languages. Get the weights at Hugging Face and AIKosh. Thanks to the good folks at SGLang for day 0 support, vLLM support coming soon. Links, benchmark scores, examples, and more in our blog - sarvam.ai/blogs/sarvam-3…
English
208
1.3K
6.9K
724.5K
Arnav Jaitly
Arnav Jaitly@arnitly·
This is the Indian Cricket Team we were promised at the start of the World Cup! Absolute madness at Wankhede. #T20WC2026
English
0
0
0
51
Arnav Jaitly retweetledi
neha
neha@neharedy·
watching countries go to war while charles and alexandra drive off in a stunning ferrari, as i open claude to work while seeing anthropic spar with the government, all while trying to hit my protein intake, feels like a very interesting time to be alive.
English
18
297
2.8K
69.3K
Arnav Jaitly retweetledi
Yash Lahoti
Yash Lahoti@YvLahoti·
Sitting in India right now, watching the world burn in real time. US & Israel striking Iran. Iran retaliating across the Gulf. Russia-Ukraine still raging. Pakistan-Afghanistan tensions simmering. Missiles flying over Tehran, Bahrain, Abu Dhabi, Qatar. And here I am, safe at home on a Saturday evening, in a democratic, peaceful country, worrying about none of this at my doorstep. We don't say this enough - we are incredibly lucky. Not perfect, not without problems. But at peace. That's not a small thing in 2026.
English
740
3.4K
27.1K
1.2M
Arnav Jaitly retweetledi
Ilya Sutskever
Ilya Sutskever@ilyasut·
It’s extremely good that Anthropic has not backed down, and it’s siginficant that OpenAI has taken a similar stance. In the future, there will be much more challenging situations of this nature, and it will be critical for the relevant leaders to rise up to the occasion, for fierce competitors to put their differences aside. Good to see that happen today.
English
1.4K
2.5K
25.6K
3M
Tanay Kothari
Tanay Kothari@tankots·
We will give you a Porsche GT 3 RS if you can type faster than @WisprFlow can dictate. Last week, we challenged 5 users to get Wispr to make a mistake. 3.5 Million people watched the challenge and wanted in. Now we're opening the challenge to everyone. Comment "Porsche" and you'll get a link to participate. Prizes apart from the Porsche: 1. Lifetime Wispr Flow Pro membership 2. 6 months of Flow Pro if you QRT with your score 3. Flow Desktop Mic 4. Exclusive Flow Merch
Tanay Kothari@tankots

We offered 5 people a Porsche 911 GT3 RS if they could get @WisprFlow to make a mistake It's the fastest and most accurate AI voice dictation app that's 3x more accurate than ChatGPT, Claude, or Siri. Today, we’re finally launching on Android. Download now: play.google.com/store/apps/det… As a part of the launch, we’re giving away 6 months of Wispr Flow Pro for free. Like, retweet and comment ‘Wispr Flow’ to get it. Enjoy. — Written with Wispr Flow

English
1.3K
183
1.3K
954K
Arnav Jaitly retweetledi
Rutuja
Rutuja@rutujax·
if you’re building in AI, it’s important to have tech friends and normal friends tech friends: - will talk claude code, openclaw, ai agents, mac mini - inspire you to grind harder normal friends: - remind you the rest of the world doesn’t give a shit about any of the above
English
42
27
418
32.2K
Arnav Jaitly retweetledi
Seb Johnson
Seb Johnson@SebJohnsonUK·
Just strolled through Soho and I must have seen close to two thousand people outside pubs drinking pints. It's 2pm on a Wednesday. Give Londoners a slither of sunshine and they are straight to the pub. No city in the world comes close to a Sunny London.
English
85
307
7.4K
207.2K
Arnav Jaitly
Arnav Jaitly@arnitly·
@ShubhAgrawal26 Let's hear it. This will give us great insights into how someone new to this would be interacting with these tools.
English
0
0
1
291
Shubh Agrawal
Shubh Agrawal@ShubhAgrawal26·
I’ve been extremely non-technical all my life and have just recently started to automate parts of my job as a growth guy for a b2b saas using claude code. Do you guys want me to share random shit I learned incase its useful for you ? Developers will consider it extremely basic, but might be useful for other people
English
42
0
217
19.6K
Harveen Singh Chadha
Harveen Singh Chadha@HarveenChadha·
10 months back parents were not happy when I left MS Today when I reached, they were smiling Dad showed me all the news channel recordings, newspapers mentions of sarvam Mom told me how she promoted sarvam in whatsapp groups and to neighbours Overall, a very small win but a long way to go
English
100
266
3.4K
70.7K
Arnav Jaitly
Arnav Jaitly@arnitly·
@Kaynvas What prompt did you use? Can't wait to try this out
English
0
0
1
25
Karthika
Karthika@Kaynvas·
I gave Claude my resume, the story behind my career, and creative direction. It produced a 60-second video resume with generative music and motion design. Honestly, I had so much fun prompting this Why do things the old way anymore?
English
245
95
2K
124.1K