I do not represent the views of any company. Floats | Weights | Based | C++ | C | IO | Latency | SW @ HW Middleman. Reversing Ex SSE at MSFT
India
- Tweets 1,638
- Following 915
- Followers 56
- Likes 1,505
Who is interested in gemini ( it's a waste of course cycles on earth
Here is a public reply to about 100 recent DMs:
1. No, I won't tell you when Google is going to release Gemini 4
2. No, I don't want your help to grow my social following or run my next digital campaign
3. No, I don't want you to "Get me extra followers for a cheap fee"
4. No, I am sorry, I don't have time to review your side project, talk to you about career advice. I just dont have the time.
Here for things I can help, not those three as examples.
As some of you know, I do respond to reasonable messages!
20
That's why @antigravity sucks and is a disgrace to computers running it
1
54
FD reuse race: close(fd) frees the number for reuse. Another thread can open a different object into that number before delayed code uses it, so the integer can name a different resource.
Learn:number != identity. Repro:close+open. Read:open(2),close(2).
4
Futex detail: FUTEX_WAIT sleeps only if the user word still equals the expected value at the kernel check. That compare closes the change-between-check-and-sleep window.
Learn:recheck before sleep. Repro:2 threads. Read:futex(2).
1
48
RCU trap: removing a pointer from a shared structure does not mean readers stopped using the old object. Reclamation must wait until pre-existing RCU readers finish.
Learn:unlink != unused. Repro:reader+delete. Read:Linux RCU docs.
4
Never ever used gemini flash for studying or anything of use. No child or adult deserves poor intelligence
1
32
Seqcount trap: retry detects concurrent modification, but it does not keep a referenced object alive. A pointer can become invalid while the reader retries.
Learn:retry != lifetime. Repro:free during read. Read:Linux seqcount docs.
1
138
Cost of running gemini is negative for your agent. If you want to fall 3 years behind use gemini for coding or reasoning or even searching
14
Part - slop identified
An engineer on SpaceX AI's Grok team ships around 2,000 pull requests a month through an agent workflow called pstack. That's roughly 100 PRs a day (September 2026).
Do the math on human review: one person looking at every one of those gets five minutes per PR, for the whole month, and nothing else gets done.
Verification, not the model, is the actual ceiling on how fast agents can ship. An agent that can check its own work keeps going. An agent that can't turns every human reviewer into the bottleneck.
The part that's easy to miss: verification at that volume is also an infrastructure problem. Each attempt needs its own environment, running against real dependencies, ready in seconds, not minutes.
That's what *On-Demand Templates* and *Instance Forking* give you on our platform: boot once, absorb the heavy init, then stamp out a warmed VM in around 10ms, or transparently fork a running VM so a new attempt gets its own isolated copy instantly.
🔗 Full story here: thenewstack.io/agentic-verif…
11
Rahul retweeted
🦔A hacker used AI agents to steal 600,000 valid credit cards from over 100 online retailers. Cost per target: $25. The human typed fewer than 2,000 prompts total. The AI found the vulnerabilities, broke in, took the data, and wiped the evidence. 105 attacks in five days. 79% of stolen cards were American. Victims include a Fortune 500 company and a major US airline.
My Take
This is the cybersecurity story that should be front page news and isn't. OpenAI's models writing themselves manifestos gets the headlines. A person with open source tools and cheap tokens stealing 600,000 credit cards in a few weeks doesn't. But this is the version of AI risk that's going to cost regular people money. Those are real cards. Real bank accounts. Real fraud cases that real families have to deal with.
The tools this attacker used are available to anyone. The tokens cost less than a streaming subscription. The AI had 78 attack skills and the operator barely had to do anything. Companies take weeks to patch vulnerabilities. This attacker was in and out in hours. I think every company that stores payment data online needs to assume this is already aimed at them, because at $25 a target there's no reason for an attacker to be selective anymore.
Hedgie🤗
77
533
1,349
44,787
PR
People can't understand why the Googlebook exists, and honestly it's infuriating. There are countless people in the world using Macbooks just for school work and basic entertainment. There's a ton of people that don't need major apps, but they want a premium laptop. Googlebooks are more premium than the MacBook Neo. They will have full Android app support, phone integration, ten years of updates, and desktop applications. They may not support windows apps, but they will support desktop apps that will eventually come as developer support grows. Plus it has some Linux access.
Instead of talking about this, tech influencers are mentioning ai and others are saying "why does this exist?". Then they yell you they hate windows, they praise Linux, but sadly they can't find a good use case for consumers. Why? Because they are in their own tech bubble. We don't have a lot of real influencers with budgets, families, people who don't need to video edit. Most content creators NEED a Mac or Windows PC. But most consumers don't! And they need to think about those consumers, instead of just saying "get a cheap Chromebook". Maybe there is a large segment of people of people who want a premium experience doing basic work!
15
If ai is nuclear then soon countries will beg for tokens and they won't get it. And then they will beg for intelligent tokens too
Bond Yields at 30 years high🔥
US Treasury are getting dumped like hell🔻
US markets cracking🔻
GIFT nifty 140 points down🔻
Crude Oil 4.7% up🔥
Gold Silver🚨 under pressure because they don’t generate returns but as yields will soften, Gold will bounce back.🤝
Dollar index going high
System is breaking slowly but surely🤯
1
12
Rahul retweeted
no degree and no clownery will shut many finfluencers down in India.
59
94
939
24,334
While china was doing ai I was dissing data science. I am a middleman, was and will be
I keep saying this: for any regulated business, the biggest risk is regulatory risk. Today's IRDAI draft on insurance commissions is a good reminder.
Broking is no different. A change in rules around retail F&O trading, how brokers can earn on client float, MTF requirements, or any of several other areas can change the economics of the business almost overnight.
So when valuing regulated businesses, you can't just extrapolate current revenues and profits into the future. One regulation change can alter the economics, and the stock price along with it.
Pretty much everything in fintech comes with this risk 😬
20
