Saurav Agarwal | Generative AI @NVIDIA

AI Architecture leader,
Enterprise GenAI playbooks, NeMo/NIM, Agentic AI
DMs open.
Weekly blueprints + slides

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

The most powerful thing about Generative AI isn’t what it creates, it’s how it changes the way we think about creativity itself.

Humans provide the spark, AI provides the scale.

Together, the possibilities are endless.

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

AI is rewiring the world: citizen bioscience, open-source agents, AI teams & Sovereign AI.
The new key skill?
AI Orchestration.

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

7/7 Lesson: Businesses pay fast for boring problems solved. Stop overbuilding. Start with niche pain + quick MVP + bold outreach. Want the exact workflow + scripts? Like/RT + reply “BLUEPRINT” DMing it free

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

1/7
People are literally selling custom AI agent for $5,000 in ONE day. No agency. No team. Just me, no-code tools, and a hungry niche. Full blueprint below 👇

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

6/7 Step 4: Price on value. Charged based on ROI: “This books you 5-10 extra appointments/month = $40k+ revenue.” $5k felt like a steal. No hourly nonsense. One-time build + 30 days support.

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

LLM hallucinations hit new lows in 2026 models, spouting confident BS like never before. Can you trust a single word from AI anymore? 🤔 #AI #LLM

[Machina @EXM7777]

how to create your own course for any tool in 2hrs:

step 1: ask Gemini to deep search all educational yt videos about the tool, it'll find the best tutorials and explanations that are already out there

step 2: save all the official documentation from the company's website, this gives you the technical baseline and ensures you're learning the correct information

step 3: search relevant subreddits to find power user tips and real-world use cases, this is where the actual practical knowledge lives that you won't find in official docs

step 4: feed everything into NotebookLM and prompt it to create a detailed course from beginner to expert with practical exercises

what you end up with is a personalized curriculum built from the best resources on the internet, structured specifically for how YOU learn, not some generic course that tries to work for everyone

the exercises it creates are based on real use cases and actual best practices, not theoretical examples that don't translate to real work

and when you get stuck, you just ask questions and the AI already has all the context from everything you fed it, so the answers are specific and relevant to exactly what you're trying to learn

this is how you compress months of scattered learning into a focused curriculum that gets you competent fast

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

@nohaahamed641 AI

[Boris Cherny @bcherny]

I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit.

My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently.

So, here goes.

[Paul Sims @SimslearnAi]

This AI Agent does full-stack SEO 🤯

Built in n8n:
📊 Analyzes GA4 + Rank + SERP
🧠 Crawls + Cleans FAQ
🔍 Tracks Competitor Keywords
✍️ Auto-rewrites articles
📈 Saves reports & performance

🔁 Like + RT
✅ Reply “AI”
🤝 Follow me & I’ll DM you the full workflow FREE

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

@Marcilenyaraa n8n

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

@SimslearnAi AI

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

KV cache pain, solved.
NVIDIA Dynamo now offloads KV cache from GPU → CPU/SSD/remote storage via NIXL—longer context, higher concurrency, lower cost.
Works with TensorRT-LLM, vLLM (SGLang coming), plus LMCache.
Reply “DYNAMO” for the quickstart. #NVIDIAAI #Inference

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

NVIDIA Rubin CPX → built for 1M+ token context.
Disaggregated inference meets a purpose-built context GPU: 30 PFLOPs NVFP4, 128 GB GDDR7, 3× attention accel (vs GB300 NVL72).
Pair with Rubin (gen) + Vera (CPU) for exascale long-context apps.
Reply “CPX” for a sizing cheat-sheet.

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

Build an Agentic RAG with NVIDIA Nemotron (free workshop)
What you get: open Nemotron models, LangGraph agent template, NIM endpoints, retriever + reranker, and a portable dev env—ship your own IT Helpdesk agent today.
Reply “RAGKIT” for the starter checklist. #NVIDIAAI #RAG

[Saurav Agarwal | Generative AI @NVIDIA @sauravagarwal]

NVIDIA open-sources Audio2Face 🎙️→🙂Real-time, AI-driven facial animation + lip-sync for lifelike 3D avatars—now with models, SDK, plugins (Maya/UE5), and a training framework available to developers. Reply “A2F” for quickstart + sample projects. #NVIDIAAI #Audio2Face #Gamedev