We build AI-native products that solve real problems — fintech, edtech, ancient knowledge, and developer tooling. Everything shipped fast, everything in product
One of the most meaningful outcomes of Mission Aagaman was the trust our customers and partners placed in us.
Vikram-1’s maiden flight successfully deployed Grahaa Space’s SOLARAS satellite and Skyroot’s own SCOPE technology-demonstration satellite, while carrying in-orbit experiments from Cosmoserve Space and DCUBED.
The mission also carried Cosmos Diamonds’ Cosmic Bloom diamond, along with micro-artist Ajay Kumar’s extraordinary creations—a microscopic 18K gold rocket and intricate micro-sculptures honouring Dr. Vikram Sarabhai, Dr. A.P.J. Abdul Kalam and Sir C.V. Raman. A tribute to India’s scientific legacy, carried into space alongside the mission.
Here’s to many more missions. 🚀
🚀 Congratulations to the Skyroot team on their groundbreaking achievements with the Vikram rocket! Your persistence and innovation inspire us all. Here's to more successful launches and pioneering strides in the aerospace industry. #Skyroot#VikramRocket#Innovation
Fable 5 is back.
When it was paused, my take was simple: progress matters, but trust matters more.
Good to see @AnthropicAI bring it back with updated cybersecurity safeguards and clearer fallback behavior. That is the right direction.
Frontier models are no longer just chat tools. They are entering real engineering workflows, touching codebases, debugging systems, and accelerating product work.
That kind of capability needs serious guardrails.
Guardrails are not friction.
They are infrastructure.
Excited to try Fable 5 again — and even more excited to see responsible deployment becoming part of the product story. @claudeai@ClaudeDevs
#ResponsibleAI #AgenticAI #AIEngineering #FrontierAI #BuildInPublic
Congratulations to @nvidia on releasing Nemotron 3 Ultra. @huggingface
The direction is clear: long-context, agent-ready models are becoming more practical.
Nemotron 3 Ultra is designed for complex agentic workflows, long-context reasoning, tool use, and high-stakes RAG, with support for up to 1M tokens.
That is a serious step forward.
But stronger models do not remove the need for Agent Ops.
They make it more important.
As agent runs get longer, the failure surface shifts from answer quality to execution control.
Did the agent stay on plan?
Did it use the right tool?
Did it stop at the right time?
Did it silently drift from the original objective?
Did it produce work that is actually done, not just 95% complete?
That is the production problem.
Better models reduce some failures.
But production agents still need:
- trace monitoring
- evals
- tool-call governance
- plan stability checks
- execution briefs
I learned this while running 13 AI products solo.
With one product, you can babysit agents manually.
With thirteen, you need an operating layer.
That is why I built V-Office as my internal Agent Ops layer.
Building agents is the demo.
Keeping them delivering is the business.
#NVIDIA #AgentOps #AgenticAI #AIObservability #LLMOps #ProductionAI #BuildInPublic
@nvidia #NVIDIAGTC @Gemini@GeminiApp
AI is no longer just in the cloud. It’s coming to the personal computer.
NVIDIA’s RTX Spark signals a new era where Windows PCs are purpose-built for personal AI agents — bringing serious AI performance closer to creators, engineers, founders, and builders.
For me, this is exciting because the future of AI will not only be about bigger models — it will be about making intelligence more accessible, local, and useful in everyday workflows.
AI is no longer just in the cloud. It’s coming to the personal computer.
NVIDIA’s RTX Spark signals a new era where Windows PCs are purpose-built for personal AI agents — bringing serious AI performance closer to creators, engineers, founders, and builders.
For me, this is exciting because the future of AI will not only be about bigger models — it will be about making intelligence more accessible, local, and useful in everyday workflows.
Proud to be part of the NVIDIA Inception startup program and looking forward to building next-gen AI solutions by harnessing engineering, acceleration, and imagination.
The AI revolution is becoming personal.
And we are just getting started. @nvidia@NVIDIAAI@startupindia
#NVIDIA #RTXSpark #AI #GenerativeAI #AIStartups #NVIDIAInception #Engineering #FutureOfAI
Behind every AI breakthrough, there is infrastructure powerful enough to make it real.
NVIDIA’s latest AI infrastructure and chip announcements show where the industry is heading: faster compute, smarter data centers, and systems designed for the age of AI agents.
From GPUs to CPUs to full-stack AI platforms, the next generation of AI will be built on engineering that scales intelligence itself.
As part of the NVIDIA Inception startup program, I’m excited to build at this frontier — creating next-gen AI solutions that harness engineering to solve real-world problems.
We build today, so the world can innovate tomorrow.
@nvidia@NVIDIAAI@startupindia
#NVIDIA #AIInfrastructure #AIChips #DataCenters #AgenticAI #NVIDIAInception #Startups #DeepTech #EngineeringInnovation
Excited to share that I've been shortlisted for the OpenAI x Outskill AI Builders Hackathon 🚀
Over the next week, I'll be building alongside 1000 selected builders from across the country to launch real AI products using Codex by OpenAI.
We'll also get access to live sessions, mentorship, feedback, and support from the OpenAI + Outskill teams throughout the hackathon.
I'll be building HireGEN — an India-first, skill-validated hiring platform that cuts through resume noise with verifiable proof of capability. AI parses GitHub portfolios, projects, and candidate evidence to build structured skill graphs. Recruiters see candidates ranked by real signal, not pedigree. The goal is to unlock Tier-2/3 talent that often stays invisible on network-biased platforms.
Really looking forward to turning this idea into a live product in just a few days while building alongside some incredible developers, founders, and AI-native builders.
I'll also be sharing my progress, build journey, learnings, and updates from the hackathon here over the next week, so do follow along and support.
If you're a developer, AI engineer, founder, or builder experimenting with AI and coding agents, you should definitely give Codex a try.
Top 5 winners receive Codex Pro access for 1 year worth ₹1.2L+.
Let's build.
#OpenAI #Codex #BuildWithCodex #AIBuilders #Hackathon #BuildInPublic
For MBA / law / public policy students —
Arthashastra is the original strategy text. Predates
Porter by 2,300 years. Covers negotiation, leadership,
governance, ethics.
VidyaSutra applies it to modern business / legal /
policy questions. Free to try.
#MBAprep#CAT2026 #LawStudents #PublicPolicy
Today, VidyaSutra is live.
An AI advisor rooted in Kautilya's Arthashastra.
You ask. It maps your situation to a 2,300-year-old strategic principle.
Cites the exact verse. Never invents one.
Ancient Intelligence. Modern Application.
🪙 https://t.co/zZ8vWfP3Jl
https://t.co/R822lPSoVL
#UPSC #IAS #UPSCprep #Arthashastra
For UPSC aspirants — Arthashastra shows up in
GS-1 (Indian Heritage), GS-4 (Ethics & Kautilya),
and the Essay paper.
VidyaSutra has all 496 verses indexed by Book +
Chapter + Verse. Ask any question, get the source
verse cited. Free tier = 5 questions/day.
@drishti_ias@civilsdaily@InsightsIAS
#UPSC #IAS #UPSCprep