
Aneesha Bakharia
@aneesha
Followers
2K
Following
52K
Media
787
Statuses
54K
problems aren’t solved by formulas; they’re solved by thinking. > this book is the timeless guide to structured problem-solving, used by mathematicians, engineers, and innovators for decades. > it teaches you how to think your way through anything
7
92
600
New AirBnb paper explains a simple loop that keeps a support LLM improving by learning from every human customer support agent interaction. Turns live customer-support work into continuous, low-friction training data. Static models go out of date as policies and products
12
47
312
My notes on nanochat, including links to the training data it uses https://t.co/rYDylb2P1c
Excited to release new repo: nanochat! (it's among the most unhinged I've written). Unlike my earlier similar repo nanoGPT which only covered pretraining, nanochat is a minimal, from scratch, full-stack training/inference pipeline of a simple ChatGPT clone in a single,
9
44
525
Document-to-Markdown converter for LLM pipelines – MarkItDown from @Microsoft This Python tool converts dozens of file types to clean Markdown, keeping headings, lists, tables, links, and metadata. Supports: - PDF, Word, Excel, PowerPoint - HTML, CSV, JSON, XML - Images (OCR +
18
202
2K
"The Impact of Artificial Intelligence on Human Thought" A big 132 page report. AI is shifting real thinking work onto external systems, which boosts convenience but can weaken the effort that builds understanding and judgment, A pattern the paper frames through cognitive
23
83
420
🚀 Introducing AI Workflow Builder (Beta) - Turn prompts into living workflows. Generate nodes, logic, and structure from text, then shape and ship your vision faster. Rolling out this week to n8n Cloud (Trial, Starter, Pro). Update to v.1.116.0 to try it. Learn more:
77
192
2K
Next week we're open sourcing Agent Builder - a visual n8n-style workflow builder for AI agents. Build AI agent workflows with a visual canvas by connecting @firecrawl_dev, LLMs, logic nodes, and MCPs, then deploy as an API. Stay tuned for this one 👀
59
133
2K
NotebookLM updates: - Infographic artefact will have a format configuration - Nano Banana will power New Video Overview formats - We will get the Fast Research option for sourcing - Magic View is no longer a mystery 🔥
19
79
824
It's easier than ever to train, optimize and run your own models thanks to open-source (versus delegating all learning, control, capabilities to black-box APIs). Cool to see @karpathy proving it once more by leveraging @huggingface fineweb ( https://t.co/WDYE0u6jPd)!
huggingface.co
Excited to release new repo: nanochat! (it's among the most unhinged I've written). Unlike my earlier similar repo nanoGPT which only covered pretraining, nanochat is a minimal, from scratch, full-stack training/inference pipeline of a simple ChatGPT clone in a single,
14
33
564
THIS chart is the CLEAREST signal of where the internet is heading. social media time is SHRINKING for the first time in HISTORY, and young people are leading the pullback. Brainrot is OUT. they grew up online, saw the full cycle of social platforms, and learned early that
590
2K
13K
As a web developer, concepts like critical rendering path, lazy loading, performance budgets, and APIs such as Navigation, Resource, User Timing, Intersection Observer, and requestAnimationFrame are essential. Explore the MDN guides on these topics. https://t.co/4lx3FltScA
2
28
202
omg testers are toast cursor already does testing in auto-pilot never touched the keyboard for this 🤯🤯🤯
39
36
574
No freaking way , nuts if these leaks are true: Gemini 3.0 can compose and play original music! https://t.co/kvodwuZiAW
120
104
1K
This survey paper argues Small language models can handle most agent tasks, and big models step in only when needed. This setup cuts cost by 10x to 30x for common tool tasks. Agent work is mostly calling tools and producing structured outputs, not recalling vast facts. So a
12
91
461
you’ll sit at your desk, talk through an idea with an AI partner, and watch it start building in real time. you’ll describe a feature and see it appear. you’ll explain a workflow and the system will connect the apis. it'll refactor your code for performance, rewrite ux copy
Codex is so good, and is going to get so amazing. I am having a hard time imagining what creating software at the end of 2026 is going to look like.
161
172
2K
framework for domain-specific knowledge extraction and reasoning with LLMs
8
133
761
I love LLMs because I can just prompt them to build simple disposable tools that I'd otherwise not - simple mermaid visualiser it even hooked it up to github pages - all with one prompt; ty codex!
5
7
87
Another crazy breakthrough Scientists at Monash University have created a tiny fluid based chip that behaves like neural pathways of the brain, potentially opening the door to a new generation of computers. A coin sized chip made from a metal organic framework (MOF) can channel
72
467
2K
“We don’t even understand how LLMs work”
New APPLE paper says a small base model plus fetched memories can act like a bigger one. With about 10% extra fetched parameters, a 160M model matches models over 2x its size. Packing all facts into fixed weights wastes memory and compute because each query needs very little.
1
2
17
So proud to see Reachy Mini named one of the Best Inventions of 2025 by @TIME! Huge credit to the @pollenrobotics and @huggingface teams, turning a concept into thousands of units sold and shipped in under 6 months. We might not be as slick as some other robotics companies (we
13
38
247