adevgir

usa

[Ruchir Bakshi @RuchirBakshi]

I’m a U.S. Army combat veteran and a Hindu American. I didn’t put on that uniform, leave my family, and go to war so some third-rate talk show guy could announce that when his side “wins,” they’re coming for my religion.

I served with Christians, Jews, Muslims, atheists, and other Hindus. Nobody pulled rank on God before we went outside the wire. Some of those men never came home. None of them died so you could fantasize about ravaging someone else’s faith. The people who worship in those temples pay taxes, run businesses, and send their sons and daughters into the same Army I served in. “When we win, we are tearing this down” is not patriotism. It is a threat against Hinduism itself to vandalize another American community’s house of worship because you hate the faith inside it.

If winning in America means demolishing Hindu temples, you don’t want the country I fought for. You just want a license to go after people who aren’t you. America is not your theocracy. Religious liberty does not expire for Hindus when you decide you’ve won. You don’t get to tear anything down. Not legally. Not morally. Not in the country for which I actually sacrificed.
@CoHNAOfficial @sudhaj12 @pushpita @carlwheless @shiva_makki @ShawnBinda @Cloudwatch199

[Marc Andreessen 🇺🇸 @pmarca]

https://t.co/lhP4mJq4Ad

[Apex Discipline @ApexDiscipline]

A very good video to educate children

[Ritesh @riteshpatel1884]

The power of repetition in tech is crazy.

1. Write code once → you understand it.
2. Write it 20 times → you remember it.
3. Build 10 projects → you start seeing patterns.
4. Debug 100 errors → you stop panicking when something breaks.

Repetition changes:
• Syntax → muscle memory
• Bugs → debugging skills
• Projects → problem-solving
• Docs → faster learning
• DSA → pattern recognition
• APIs → better system understanding
• Deployments → confidence
• Failures → better engineering decisions

Build → Break → Debug → Fix → Repeat.

[Better Men @Better_men]

YOU ARE NOT DEPRESSED. YOUR LIFE JUST HAS 0 SIDE QUESTS.

Here are 33 side quests to complete before next month ends:

[Vivek Wadhwa @wadhwa]

H-1B’s death should cheer Indians and Americans.
This is about an article I wrote telling India why it should support America terminating the H-1B visa program:
H-1B was created to fill genuine shortages of highly skilled workers. It was supposed to allow American companies to recruit exceptional people whose expertise could not be found here. Instead, body shops, in collaboration with American employers, turned it into one of the most corrupt labor-arbitrage schemes in modern America.
The body shops learned to flood the system with applications, manufacture experience, embellish résumés, and present ordinary technology jobs as positions requiring rare skills. Workers were brought in at lower wages and placed at American companies through layers of contractors and subcontractors, allowing the companies to pretend that they had not directly replaced anyone.
American workers were harmed the most. Many were ordered to train their foreign replacements before losing their jobs. Experienced engineers were pushed out and told that their skills were obsolete, while younger and cheaper workers were brought in to perform substantially the same work. Age discrimination became an unspoken part of the business model.
These were decent, hardworking Americans who had studied, built careers, supported families, and contributed to their companies for years. Their loyalty was rewarded with termination notices because executives had discovered a cheaper and more controllable source of labor.
The control was central to the scheme because an H-1B employee whose right to remain in America depended on the sponsoring company could not negotiate like an American worker. Leaving a bad job could mean losing legal status and being forced to leave the country. The decades-long green-card backlog made this dependency even worse. Employers gained workers who were less likely to demand raises, report abuse, challenge management, or resign.
Then the companies wrapped this exploitation in the language of diversity, innovation, and skilled immigration. Executives claimed there were real talent shortages while laying off qualified Americans.
The darkest part is that the corporations escaped most of the blame while Indians became the public face of the abuse. American workers saw jobs being transferred to Indian contractors, heard stories about fabricated experience and embellished résumés, and concluded that Indians were cheating them. Some applicants certainly participated knowingly, and that dishonesty must also be acknowledged. But the system was created, expanded, and defended by companies that made billions from it. This is all they lobbied for in DC, not real solutions to the backlog.
The consequences have been devastating for Indians in America. Millions of legal immigrants who studied hard, worked honestly, paid taxes, created companies, and contributed enormously are now viewed with suspicion. The achievements of an entire community are being overshadowed by unscrupulous players. The anger has spread far beyond H-1B and mutated into open anti-Indian hatred.
The visa also exploited the Indian workers it brought to America. Their legal status depended on their employers, and the decades-long green-card backlog left many unable to change jobs, start companies, or plan their lives. They were sold the American dream and then trapped in a form of indentured servitude.
That is why this cancer, the H-1B visa must be terminated, and terminated now. And this is why I am speaking up so strongly. Enough is enough.

[Patrick OShaughnessy @patrick_oshag]

Neil Movva (@neilmovva) started his career at Nvidia, working on GPUs and kernels, and has an unusually deep understanding of inference, from software to chips to power.

We spend a lot of time on each of those layers, how they connect, and where the important tradeoffs are.

What makes this conversation special is how detailed it is (like a 401-level class), yet Neil makes it remarkably clear and easy to follow.

Today he runs Sail Research, a company building infrastructure for agents to make tokens as cheap as possible.

We discuss:
- Latency versus throughput
- Why there are no bad chips, only bad pricing
- The end of kernel engineering
- Buying chips and power no one else wants
- New chip architectures
- Nvidia lore + his contrarian view of the company
- Open source and the frontier labs

I learned a ton. Enjoy!

TIMESTAMPS
0:00 Intro
0:38 Building a “Token Factory”
4:21 The Future of Background Agents
13:09 Nvidia and the GPU Stack
23:27 Chips, Memory, and Transformers
36:14 The Future of AI Training Data
44:32 Chip Scarcity and Compute Arbitrage
52:44 Reinventing the AI Data Center
59:01 Power and the “Scavenger Strategy”
1:10:10 Open vs. Closed AI

[Mark Zuckerberg @finkd]

(1) Today we're releasing Muse Spark 1.1 -- a strong agentic and coding model at a very low price. It's available through our new Meta Model API and in Meta AI.

[Anatoli Kopadze @AnatoliKopadze]

https://t.co/AAWIZD1pNL

[Vishakha Singhal @vishisinghal_]

Breaking 🤯

The entire RAG industry is about to get cooked.

Researchers have built a new RAG approach that:

- does not need a vector DB.
- does not embed data.
- involves no chunking.
- performs no similarity search.

It's called PageIndex. Instead of chunking your docs and stuffing them into pinecone, it builds a tree index and lets the LLM reason through it like a human reading a book.

hit 98.7% on financebench. beats every vector RAG on the leaderboard.

no embeddings. no chunking. no vector DB.

100% open source.

[Aparna Dhinakaran @aparnadhinak]

https://t.co/3aL1sOXBKW

[Ashpreet Bedi @ashpreetbedi]

https://t.co/suj0MIw2Cm

[Shiv @shivsakhuja]

https://t.co/F0b4Ehn8uZ

[Augment Code @augmentcode]

https://t.co/F9oAhFsnkN

[Akshay 🚀 @akshay_pachaar]

A harnessed LLM agent.

Most people picture this as a model with tools bolted on. The real architecture inverts that relationship.

The model itself is deliberately thin. Intelligence gets pushed outward, and the harness composes it at runtime.

Three dimensions orbit the harness core:

𝗠𝗲𝗺𝗼𝗿𝘆 holds state the model shouldn't carry in weights or context. Working context, semantic knowledge, episodic experience, and personalized memory each have their own lifecycle.

𝗦𝗸𝗶𝗹𝗹𝘀 hold procedural knowledge. Operational procedures, decision heuristics, and normative constraints specialize the general model per task.

𝗣𝗿𝗼𝘁𝗼𝗰𝗼𝗹𝘀 hold the interaction contracts. Agent-to-user, agent-to-agent, and agent-to-tools are three distinct surfaces with their own failure modes.

Between the core and these modules sit the mediators: sandboxing, observability, compression, evaluation, approval loops, and sub-agent orchestration. They govern how the harness reaches out and how state flows back in.

The useful question this framing unlocks: for any new capability, where should it live? Stable knowledge goes to memory, learned playbooks go to skills, communication contracts go to protocols, loop governance goes to the mediators.

Harness design becomes a question of what to externalize, and how to mediate it.

I'm building a minimal agent harness from scratch. Didactic, easy to read, no magic. Open-sourcing it soon. Stay tuned.

[Akshay 🚀 @akshay_pachaar]

https://t.co/X7JmlflF3U

[𝗿𝗮𝗺𝗮𝗸𝗿𝘂𝘀𝗵𝗻𝗮— 𝗲/𝗮𝗰𝗰 @techwith_ram]

Just finished reading this blog on agent harnesses. Man, it’s one of the clearest, most practical takes I’ve seen why they’re here to stay and why memory isn’t some optional plugin.

[Mason Daugherty @masondrxy]

If you read any part of this, check out the FAQ

[Ashwin Gopinath @ashwingop]

https://t.co/kmxV6P8c4Q

[Amit Shekhar @amitiitbhu]

My recent 10 articles on X:

- KV Cache in LLMs
- Paged Attention in LLMs
- Causal Masking in Attention
- Byte Pair Encoding in LLMs
- Harness Engineering in AI
- Math behind Attention - Q, K, and V
- Math behind √dₖ Scaling Factor in Attention
- Math Behind Backpropagation
- Decoding Transformer Architecture
- Mixture of Experts Explained

X is a knowledge sharing platform.