Sumit Gupta

[Make It With AI @makewith_AI]

SpaceX just quietly amended its S-1 announcing another mega deal

$920M/month from Google from October 2026 through June 2029

With both parties being able to terminate the agreement

with 90 days notice Things are getting exciting 🚀

[Sumit Gupta @SumitGu33113871]

@IndianTechGuide Then set an example from own company.

[Make It With AI @makewith_AI]

Most people use Claude, ChatGPT, and Gemini like a search engine.

That's why they get average results.

Stop saying:
❌ "Write code"
❌ "Fix this bug"
❌ "Teach me AI"
You're talking to a senior engineer like it's a junior intern.
The quality of your prompt determines the quality of your outcome.
Here are 8 prompts that instantly make AI 10x more useful:
🧵👇

[Make It With AI @makewith_AI]

https://t.co/NPWG7lPCjm

[Make It With AI @makewith_AI]

NVIDIA and Microsoft might be building the first PC made for the AI agent era.

right now, your laptop is just a tool.

you open apps.
you click buttons.
you do the work.
but that's starting to change.
AI can now run directly on your computer, and the new models are smart enough to handle real tasks on their own.

instead of waiting for commands, your PC can help with coding, writing, research, editing, and more.

think of it like having a personal assistant built into your laptop.

the computer doesn't replace you.

it works alongside you.

that's what people mean when they talk about the "next generation of PCs."

[Make It With AI @makewith_AI]

pov: assistant professor in cse and your students still think chatgpt is "hacking"

most cybersecurity classes are stuck teaching 2022 threats.

genai rewrote the rules in 18 months.

i'm an assistant professor exploring ai tools, security flaws and emerging tech.

building weekly breakdowns for people who actually ship.

today's signal:

one researcher used claude 4 + cursor and found a zero-click rce in under 40 minutes.

teams used to need weeks.

what actually matters right now:

1/ prompt injection is the new buffer overflow

2/ agent swarms chain vulns across 7 tools with zero human help

3/ most "ai security" tools are still just regex with llms on top

the window is open. the bar is low.

if you're in ai / genai / cybersecurity — reply with the biggest security flaw you've seen genai create this year.

i'll drop the exact prompt + reproduction steps for that rce in the replies.

professor mode on.

let's build better defenders.

[Make It With AI @makewith_AI]

India has some of the world's best engineers.

Yet we rarely build the technologies the world depends on.

Anthropic was founded in 2021 and is already worth hundreds of billions. Meanwhile, many Indian IT giants built decades ago still primarily earn revenue by providing services rather than creating breakthrough technology.

The difference isn't talent.

It's investment in research, risk-taking, and long-term innovation.

They build:
• Foundation models
• Chips
• Operating systems
• Robotics
• Deep-tech platforms
We mostly:
• Integrate
• Outsource
• Implement
• Consume
A nation doesn't become a technology superpower by using technology. It becomes one by creating technology.
India doesn't need more apps.
India needs more labs, more researchers, more patents, more deep-tech startups, and more founders willing to spend 10 years building something the world cannot ignore.
The future belongs to the countries that invent it.

[Sumit Gupta @SumitGu33113871]

@makewith_AI Amazed

[Make It With AI @makewith_AI]

Good morning everyone 🌄

[Make It With AI @makewith_AI]

The AI model race is getting absurdly fast 👀

A 1.5T parameter Grok model trained with massive coding data + upcoming RL tuning means coding agents are about to get another huge jump in capability, especially for difficult engineering tasks.

[Sumit Gupta @SumitGu33113871]

@PaulleyTicks @grok it's real?

[Sumit Gupta @SumitGu33113871]

@FarzamHejazi Claude

[Make It With AI @makewith_AI]

Gemini Omni represents a huge leap in multimodal AI and world understanding.

It can combine photos, videos, and audio to generate entirely new scenes — and this is only the beginning. Over time, it aims to support almost any kind of input and output, starting with advanced video generation and editing.

[Make It With AI @makewith_AI]

Big move for the open-source AI ecosystem 🚀

X Premium members can now connect their subscription directly with OpenClaw — an open-source, local-first AI assistant.

Search X posts, generate images/videos, chat with your agent, and build personalized workflows with more control and privacy 👀

[Make It With AI @makewith_AI]

🔥 Anthropic’s rumored “Claude Mythos” model has reportedly appeared inside Google Cloud Console 👀

It still seems unlikely that Anthropic would fully release the model publicly, but they may provide access through selected enterprise partnerships running on Google Cloud infrastructure.

If true, this could point toward a more controlled “AI-as-infrastructure” approach rather than an open public rollout.

[Sumit Gupta @SumitGu33113871]

@elonmusk That's Insane

[Sumit Gupta @SumitGu33113871]

BJP ne UK me muslim Vs christian karwa Diya

[RadioGenoa @RadioGenoa]

"Allah, Allah, who the fuck is Allah?" Little British patriots grow up.

[Sumit Gupta @SumitGu33113871]

@makewith_AI 👍

[Make It With AI @makewith_AI]

A 19-year-old student from Bihar just built a 5.82B parameter multimodal AI model after starting from absolute zero just 2.5 years ago. 🤯🇮🇳

No elite university.
No billion-dollar lab.
No Silicon Valley backing.
Just curiosity, persistence, and sleepless nights learning AI from scratch.
The most powerful part wasn’t even the model.

It was this:

“₹9.64 lakhs on GPU compute is not a small number for us.”

That sentence alone explains the sacrifice, risk, and belief behind this project.

We are entering an era where one determined person with internet access, cloud credits, and obsession can build things that once required entire corporations and research labs.

Talent is everywhere.
Opportunity is finally catching up.

Massive respect to Abhinav Anand 👏

Follow @makewith_AI for more insights into the AI revolution 🚀

[Make It With AI @makewith_AI]

The Anthropic Fellows Program is honestly one of the most interesting opportunities right now for people trying to break into serious AI research, AI safety, security, and advanced ML engineering.

What makes it different is that they’re not only looking for traditional PhDs from elite labs. They explicitly say they want strong technical builders, researchers, engineers, and security people — even if your background is unconventional.

The program offers:

• 4 months full-time research
• Direct mentorship from Anthropic researchers
• Up to ~$15k/month compute funding
• Weekly stipend (~$3,850 USD)
• Workspaces in London & Berkeley
• Remote options for US/UK/Canada
• Research tracks in:
AI Safety
AI Security
Reinforcement Learning
ML Systems
Economics & Policy
What’s especially interesting is the AI Security track.

They’re looking for people with:
pentesting experience
vulnerability research
open-source contributions
exploit development

LLM/security research

bug bounties & CVEs

offensive security backgrounds
This is a huge signal that frontier AI labs now see cybersecurity as a core part of AI development, not just an IT function.
The fact that Anthropic is funding researchers to study:
adversarial robustness
model control
mechanistic interpretability
AI red teaming

catastrophic risk reduction

AI-enabled cyber capabilities

shows how seriously the industry is starting to take AI safety and AI security.
For students and researchers in cybersecurity, NLP security, AI red teaming, RL, or ML engineering — this is the type of opportunity worth paying attention to 👀