Indrajit Bhosale | इंद्रजित भोसले

Computer Systems | History | GGMU | Running (In this particular order)

[Indrajit Bhosale | इंद्रजित भोसले @devnull_0]

My time at @CarnegieMellon @SCSatCMU and @inicmu ends :)
Special thanks to @_ritvika_ @BhosalePratim and @ShivaniRB16 for their support throughout one of the most challenging phases on my life.
I already miss the grind :)

[Mohamed @MoeSquare]

I mean this as a compliment, but Arsenal over the past ~2 yrs have morphed into the modern day version of those Mourinho-coached Chelsea sides from the mid 2000s who conceded 15 and 22 goals in 05 and 06

[Josh Hart @joshhart]

You couldn’t teach the type of Hate Watch I’m on today

[𝗜𝘃𝗮𝗻 𝗦𝗰𝗵𝘄𝗮𝗸𝗼𝗳𝗳 @Ivan_Schwakoff]

Only Game Of Thrones fans will know 😂😂

[Indrajit Bhosale | इंद्रजित भोसले @devnull_0]

It would be very tragic if I woke up at 7:30 and went through all this to see Arsenal win the double

[Indrajit Bhosale | इंद्रजित भोसले @devnull_0]

@fracturebaandya This but 5x with ABCDs

[Mitchell Hashimoto @mitchellh]

I've got an agent in a loop optimizing a renderer with the goal to minimize frame times (and tests to measure). It got times down from 88ms to 2ms and allocations down from ~150K to 500. Sounds good, right? Wrong. This is exactly why agent psychosis is a big fucking problem.

As an experiment, I rewrote the Ghostty core render state in Go, with access to identically laid out data structures as Ghostty and the exact same validation tests. I made a purposely naive renderer (simple, correct, but slow). 88ms per frame with 150,000 allocations (horrendous, lol)!

I then kickstarted a Ralph loop to bring the frame times down. I told it it can't modify input data structures or the public API or tests (they're correct), but it can do anything else it wants. It got to work.

It has worked for about 4 hours. I've spent around $350 on this experiment so far. The results?

88ms => 1.5ms
150K allocs => ~500 allocs

Incredible right? Nope.

My hand-written renderer I ported has frame times (same benchmark) of ~20us (0.020ms) and 0 allocations in the update path.

This is the problem with psychosis and lacking systems understanding. If you don't understand the system, you're going to accept that this is an incredible result. If you understand the system, you'll see better solutions immediately and can do roughly 75x better on throughput.

The people who blindly trust agent output are in the former camp. They're sheeple, overdrinking from a fountain of mediocrity.

Standard disclaimer: I use AI all the time. I like AI. The point I'm making is to not blindly accept results. Think. Analyze. Learn.

[Arjun* @mxtaverse]

CBSE reportedly gave contract to the exact same tech firm (with name changed) that did a massive fuckup in Telangana in 2019 where 20 students committed suicide and 300k+ students faced discrepancies.

[Anmol @Anmol_299]

I am no expert in this field but there is a genuine question......Why is everything in white?

From when did Maharashtrian Temples start relying so much on White Marbles

Traditional Maharashtrian Temple Architecture doesn't look like this, this looks like Iskcon or Jain Temple

[Shilpa Godbole @godbole_shilpa]

This sounds patriotic & appealing in theory, but is India actually ready for it?
I can't speak for every metro city, but I can certainly speak for Pune. Our basic infrastructure is struggling to keep up with reality. Public transport isn't expanding at the pace the city's population is growing. The Metro is moving forward, but painfully slowly. The traffic situation is already in shambles.
10 or 15 yrs ago, I genuinely believed Pune could become a great city for Indians returning from the US or UK to settle down in. Today I don’t have that optimism. The apathy shown towards Pune by those in power has steadily eroded that confidence.
Before we dream of reverse migration on a large scale, we need cities that can actually support the people already living in them.

[newfriend @Cereal_fan]

Laid off NRI - omg what am I going to do now ? I have a wife on h4, school going kids, and just 10 properties in prime Mumbai and Pune locations 😭
Plz help me Modizeee 😫

[unusual_whales @unusual_whales]

BREAKING Nvidia, $NVDA, earnings:

- EPS: $1.87, est: $1.77
- Revenue: $81.6 billion, est: $78 billion

[झुंझारराव @Ranzunjhar]

The people of Bengal themselves regard Shivaji as a hero; even Subhas Chandra Bose considered him an ideal. So yes, Shivaji has much to do with Bengal. Shivaji belongs to everyone.

[Indrajit Bhosale | इंद्रजित भोसले @devnull_0]

Raja Shivaji is surprisingly decent.
First half drags, second half delivers. Music alone is worth the ticket. 10/10 Goosebumps.
Finally, a pan-India film celebrating a pan-India hero.
Could it have been better? YES
Is it worth watching once? Absolutely YES
Am I biased? YES

[The Analyzer (News Updates🗞️) @Indian_Analyzer]

Devendra Fadnavis:
🗣️ “Whoever lives in Maharashtra should try to learn MARATHI.
— If they haven’t learnt it, we will teach them, but VIOLENCE is not acceptable."

From today, it is MANDATORY for all rickshaw drivers to know Marathi.

[Sachin Tendulkar @sachin_rt]

अभिमान आहे मराठी असल्याचा,
माझ्या मातीचा, माझ्या महाराष्ट्राचा,
महाराष्ट्र दिनाच्या हार्दिक शुभेच्छा!

[Indrajit Bhosale | इंद्रजित भोसले @devnull_0]

@sachaarbonel @theCTO Checkout Brev it's more than a sandbox but I use it as a sandbox.

[Indrajit Bhosale | इंद्रजित भोसले @devnull_0]

@jukan05 @maheshbapatu

[Jukan @jukan05]

Thoughts after reading the DeepSeek V4 paper:

- NVIDIA really is something else. Remember how back in 2024 people were bashing Blackwell as overspec'd and dismissing FP4 as just marketing? Turns out it was all groundwork for the next generation of models. Maybe NVIDIA's moat is its ability to anticipate the trajectory of mainstream LLM technology and the new demands on accelerators 3–5 years out, plan and position accordingly, and bake that foresight into product design. Other GPU companies don't anticipate demand — they react to it.

- Have NVIDIA and DeepSeek been talking to each other? Looking at the 6144 FLOPs/Byte passage — I'd been wondering why NVIDIA was pushing HBM4 pin speeds up so aggressively, and it turns out raising Rubin's HBM4 pin speed isn't "overkill" from the perspective of a model like V4. It's a precisely balanced design.

- NVIDIA is once again working hard to crank up bandwidth on Rubin Ultra, which is telling: it implies that Rubin Ultra's FP4 compute is getting too fast relative to HBM bandwidth, and that HBM bandwidth could once again become the bottleneck when training MoE models like DeepSeek-V4.

- Why is the next-gen NVIDIA chip scaling up the NVL domain? Why are they moving toward Kyber? You can read this as an attempt to push up the bandwidth-density of the interconnect fabric, pulling compute — which has surged above the threshold — back into a communication-friendly balance.

- The upshot is that the DeepSeek paper is essentially telling us NVIDIA's current chip design lines up with DeepSeek's model patterns. If Blackwell alone produced this much evolution, imagine what Rubin and Feynman will deliver.

- NVIDIA's G3.5, unveiled this year, is almost uncanny. It's a new tier sitting between local NVMe SSDs and object/file storage — one that exists only for AI inference — meaning NVIDIA has created an entirely new memory-tier category exclusively for AI workloads. And in §3.6.2 of the V4 paper, DeepSeek argues that the KV cache can break out of GPU HBM's limits and be permanently offloaded to NVMe storage. That maps exactly onto the ICMS rack NVIDIA showed at CES. NVIDIA called it precisely — they saw that labs like DeepSeek would need this.

[Indrajit Bhosale | इंद्रजित भोसले @devnull_0]

Why do all new BMW suv's look like an EV but actually aren't.
Such a self goal from the company. If anything you'd want to stand apart!
This is the BMW X3 (Again Not an EV)

[Beff (e/acc) @beffjezos]

So, given the Dwarkesh interview, Mythos was really a natsec-maxxing psyop to push the admin to stop selling GPUs to China so that they can temporarily slow down the non-stop distillation assault on Claude by Chinese companies.

Got it.