One of the best recent podcasts on AI. Neil has a gift for explaining all of the jargon and insights simply.
(I’m not involved, just found it unusually educational).
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
shimba retweeted
Prime Intellect engineer:
"everyone's bragging about a million-token context. here's what they don't tell you.
at 256k tokens GPT-5.5 scores 80% on retrieval. push it to a million and it drops to 36%. the model accepts the context, it just can't reason across it. people call it context rot."
in a 20-minute talk he explains why bigger context windows won't save your agents.
continual learning + training on your own traces + real environments - that's the fix.
Watch the talk, then save!
Andrew Ng just dropped a free course on Claude Code from scratch, taught with the Anthropic team:
00:00 - why Claude Code is so agentic
04:00 - shockingly simple architecture
12:00 - point it at any codebase
this short watch will replace 10 paid coding agent courses.
Andrew Ng calls it his personal favorite coding assistant right now.
Watch it today, then read how to engineer your own agent loops in the article below
shimba retweeted
If you want to actually understand memory allocation, read this by @TheGingerBill.
A series so damn good.
gingerbill.org/series/memory…
Pure Gold
Atlassian's revenue: $1.79 billion last quarter
Atlassian's move: fire the engineer who built their infrastructure
his move: post a 38-minute breakdown of every system he built, free for anyone to copy
what he revealed:
> Envoy proxy instead of enterprise load balancers
> sidecar architecture for auth, logging, rate limits
> DynamoDB + SQS for async provisioning
> Packer + SaltStack for automated VM deployments at scale
Atlassian charges per employee across 350,000 customers
this guy just handed you the enterprise playbook for free
save this
83
shimba retweeted
🚨 THE DOMINOES ARE FALLING. IN DAYS, NOT WEEKS.
Here is the global oil crisis — country by country — as of today:
🇱🇰 Sri Lanka — RATIONING. Nationwide QR-code fuel system for cars & motorcycles. Public offices & schools/universities on conservation schedule.
🇵🇰 Pakistan — CRISIS. 4-day work week for public & private sectors. Unprecedented price surge. Remote work pushed & markets closing early.
🇮🇳 India — VULNERABLE. SPR ~6–10 days left. Total stocks ~60 days cover but panic growing. Govt scrambling for emergency imports.
🇰🇷 South Korea — CLOCK TICKING. Odd-even driving rules (mandatory for public sector, voluntary for private with discounts). Fuel price caps in effect.
🇯🇵 Japan — RELEASING. Claimed 230+ days. Extra strategic reserves now being tapped (second major release started May 1).
🇬🇧 UK — PRICE SHOCK. Targeted aid packages for heating-oil users. Windfall tax talks & anti-gouging measures active.
🇩🇪 Germany — TAX RELIEF. Petrol & diesel tax cut extended. Employer fuel bonuses rolled out.
🇫🇷 France — AID LAUNCHED. Targeted fuel discounts & allowances for high-mileage drivers, transport, fishing & agriculture. Energy vouchers accelerated.
🇿🇦 South Africa — LEVY EXTENDED. Fuel levy slashed to ease pump prices. Citizens still reporting pressure & queues at stations.
🇹🇷 Turkey — TAX CUT. Special consumption tax on fuel lowered to cushion consumers.
🇧🇷 Brazil — SUBSIDIES. Diesel taxes scrapped & subsidies for producers/importers in place.
🇦🇺 Australia — VOLUNTARY CUTS. “Every Little Bit Helps” campaign + excise duty halved. Business support loans active.
🇺🇸 USA — SPR DRAWDOWN. Part of IEA’s largest-ever coordinated strategic release. State-level gas tax relief active.
🇨🇳 China — STOCKPILED & LOCKED. Massive reserves held back. Refined-oil export ban & domestic price controls still enforced.
This is DAY 70 of the Hormuz blockade.
20% of global oil — GONE.
8 million barrels per day — GONE.
IEA’s response: largest-ever strategic stock release in history. OPEC+ increases remain minimal relative to the hole.
Here is what nobody is telling you:
They’re showing you “reserves are sufficient.”
They’re NOT showing you India’s SPR at ~6–10 days, Pakistan on 4-day weeks with rationing, Japan quietly tapping extra reserves, or Sri Lanka’s QR-code fuel system. Empty pumps and demand destruction are already here in Asia.
The real energy crisis hasn’t even started yet.
shimba retweeted
Last week I gave a talk at AI Dev ’26 by @DeepLearningAI on “AI can’t read PDFs, how do we fix it” .
I’m sharing the slides publicly if others are interested in doing a deep dive into document understanding.
AI agents are going to automate huge amounts of knowledge work, but knowledge work depends on data, a lot of that data is in documents/PDFs, and existing OCR tools suck.
PDFs are a format that is inherently hard to read, and I dive into specific reasons why (tables, layouts), and why frontier VLMs and benchmarks are still insufficient.
Even as agents get better and more general, they need the right tools to read and act over PDFs. They both need this at the data ingest layer, as well as tools they can call on the fly.
Check out the slides: figma.com/deck/v4xhu6Q797nLN…
We’re building high-quality AI document processing, both with LlamaParse, along with OSS efforts like LiteParse and ParseBench.
If you have a ton of PDFs that you’re hoping to unlock with AI, come talk to us! llamaindex.ai/contact?utm_so…
shimba retweeted
Heard on TV
Whether the Elephants fight or make love, the grass always gets crushed.
India mkts is the grass
shimba retweeted
Nice blog from Kailash, CTO of Zerodha
TLDR- An experienced developer who can talk well, that is, imagine, articulate, define problem statements, architect and engineer, has a massive advantage over someone who cannot, more disproportionately than ever nadh.in/blog/code-is-cheap/
Where the hell you get the time to go through all these @aravind
Replying to @aravind
No, this is not BS on nutrition. It is not against protein as you are trying to always make it out to be. What I am saying is very sound advise on moderation. Also let me add, the development of diabetes depends on the type of protein. Animal protein increases the risk of type 2 DB. Vegetable protein with fiber reduces it. A very good quality research paper ( systematic review + meta-analysis) shows that total protein and animal protein could increase the risk of T2DM in both males and females, and plant protein decreases the risk of T2DM in females. pubmed.ncbi.nlm.nih.gov/2887…
43
shimba retweeted
Old timers got it right. And when you're poor, you don't have much choice. So our parents, their parents, did what they thought best back in the day.
Bought their house the hard way - slogging, saving, without home loan, borrowing from friends and family. Bought gold because gold is gold. Didn't understand stock market so stayed away. Bought an insurance policy because LIC is LIC and you get bonus.
Look what it got them.
Their kids got a home. That's the best security in a poor country. No home loan. They got the LIC policy. They got gold. All great products worth security and when you're poor, security means a lot.
Sure they missed the Infosys IPO but they also missed so many scams.
What did they do? They sent their kids to schools, colleges, and in a liberalised open India, these kids became VP, senior VP, in banks or IT firms and made more money than their parents and bought even bigger homes. And got very lucky with ESOPs. This is what education and economic reforms got for them.
Now their kids. Gen Z. With the security of parents who worked in MNCs, in the comfort of their 4BHK in Lower Parel. They write long posts on social media about how old India didn't have choice and invested in "garbage" like LIC policies, FD, gold, etc. Whereas the cool way is to log on to an app and invest in an index fund.
Sharam karo saalon. Your grandparents and your parents saw what the real India was. Have some respect.
For likes and follows you're ignoring and forgetting what India used to be and how prior generations gave their best and came out with results they're proud of.
Sharam karo.
shimba retweeted
🚨 BREAKING: Google Research just dropped the textbook killer.
Its called "Learn Your Way" and it uses LearnLM to transform any PDF into 5 personalized learning formats. Students using it scored 78% vs 67% on retention tests.
The education revolution is here.
shimba retweeted
Andrej Karpathy says you should learn AI depthwise, not breadthwise.
Most education is breadthwise: watch lectures, memorize formulas, and trust you'll need it later.
Karpathy flips this by learning "depthwise, on demand."
What this means:
Pick a project, start building, and learn exactly when you hit a wall.
When he created a tutorial on transformers (the architecture behind ChatGPT), he didn't start by explaining attention mechanisms or complex architectures.
Instead, he started with the simplest possible thing: a lookup table that predicts the next word.
You build that first. Then you try to make it handle more complex patterns.
And it breaks.
Only then, when you've felt the limitation, does he introduce the next concept. Each piece solves a problem you've actually encountered.
As he puts it:
"It's a dick move to present the solution before I give you a shot to try it yourself."
When you attempt the problem first, the solution actually makes sense.
Teaching forces you to learn. "If I don't really understand something, I can't explain it."
When you try to explain and stumble, you've found the gaps in your understanding.
...
Build a project that gives you a reward.
Hit a wall. Learn just enough to solve it. Then explain it to someone else.
Don't consume content. Build the code.
That's how you actually learn.
shimba retweeted
Everyone wants to get into robotics.
No one knows where to start.
LeRobot's Francesco just dropped a 70-page crash course that takes you from zero to cutting-edge:
- RL sim/real
- ACT, Diffusion policies
- VLAs, SmolVLA, Pi-0
Absolute gold if you want to catch up fast.
A comprehensive, hands-on tutorial on the most recent advancements in robotics 🤟
...with self-contained explanations of modern techniques for end-to-end robot learning & ready-to-use code examples using @LeRobotHF and @huggingface. Now available everywhere! 🤗



































