The Best AI Engineers to Follow on X in 2026

If I wanted a clean AI feed on X in 2026, I’d start with 10 names and sort them into 3 jobs: news, building, and research. That’s the core idea here.

This list points me to the people most likely to help with:

• frontier AI updates from OpenAI, Anthropic, Meta, and more
• hands-on engineering for agents, RAG, APIs, tooling, and deployment
• research context that keeps hype in check

The article names 16 accounts in total, but a few stand out by use case:

• Sam Altman, Dario Amodei, Andrew Ng for model direction and lab updates
• Andrej Karpathy, Harrison Chase, Jerry Liu for code, agents, and RAG systems
• John Carmack, Simon Willison, Vin Vashishta for systems, security, and shipping models
• Yann LeCun, Lex Fridman, Kai-Fu Lee for research debate and market context

The short version: if you build with AI, don’t follow “AI commentary.” Follow people close to the work.

Quick Comparison

Account	Main use	What I’d watch for
@sama	Frontier news	OpenAI direction, release hints
@karpathy	LLM engineering	Training, code, agent workflows
@ylecun	Research debate	LLM limits, world models, open models
@AndrewYNg	Applied AI	Agent use, implementation ideas
@darioamodei	Scaling and safety	Lab thinking, risk, model progress
@alexalbert__	API usage	Prompt caching, token flow, API tips
@AmandaAskell__	Model behavior	RLHF, character tuning
@hwchase17	Agents	LangChain patterns, memory, tools
@jerryjliu0	RAG	Indexing, retrieval, eval
@ID_AA_Carmack	Systems	Performance tradeoffs, low-level thinking
@vinvashishta	MLOps	Deployment failure points, team setup
@simonw	Security/tooling	Prompt injection, local models, CLI work
@amanrsanger	IDE workflows	Cursor routing, developer telemetry
@lexfridman	Research summaries	Interview takeaways, lab context
@kaifulee	Global AI business	Asia market movement
@AnthropicAI	Product updates	Claude and platform release news

A simple way to use this: build a private X list with 10 to 15 accounts, then turn alerts on for only 3 to 4 that affect your stack. That cuts noise fast.

I’d use X for the first signal and longer reads for the extra detail.

Best accounts for LLM and frontier AI news

If you want to keep up with model releases, API changes, and sudden jumps in capability, it helps to follow a short list of people who consistently post useful signal.

Sam Altman (@sama), Andrej Karpathy (@karpathy), and Yann LeCun (@ylecun)

Sam Altman tends to post at the big-picture level: OpenAI’s direction, AGI timing, policy stance, and short hints about upcoming models. He’s best for frontier direction and release context, not deep technical detail.

Andrej Karpathy is one of the clearest voices in engineering when it comes to how LLMs work. He breaks down model architecture, training, and how to build with LLMs instead of just around them, which makes his posts especially useful for builders. His X threads, plus the longer write-ups that show up on daily.dev, are worth reading all the way through.

Yann LeCun, Meta’s Chief AI Scientist, plays the role of a strong skeptic in a space that often runs hot. He posts daily about why he believes LLMs won’t reach AGI, why world models matter, and why open-source architectures matter. That pushback is the main reason to follow him. His feed helps balance out the scaling hype.

Andrew Ng (@AndrewYNg), Dario Amodei (@darioamodei), and Anthropic engineering voices

For a broader view of frontier AI, these accounts are also worth your time. Andrew Ng focuses on practical implementation and agentic AI. Dario Amodei doesn’t post often, but when he does, his threads on scaling and safety tend to matter.

For product and release updates, follow @AnthropicAI. For the nuts and bolts, follow Alex Albert (@alexalbert__), who shares API tactics, prompt caching tips, and token routing, and Amanda Askell (@AmandaAskell__), who writes about RLHF pipelines and Claude’s character tuning. If you care about hands-on engineering detail, this is where a lot of the useful stuff shows up.

Account	Best for	Signal frequency
@sama	Frontier direction, release context	Often brief, high-impact
@karpathy	LLM architecture, training, and building with LLMs	High-density
@ylecun	Scaling skepticism, open-source research	Daily
@DarioAmodei	Safety, alignment, scaling laws	Infrequent
@AndrewYNg	Agentic AI, practical implementation	Consistent
@alexalbert__	API tactics, prompt caching, token routing	High-density

Next, move from news to the accounts that show how to turn these model shifts into working systems.

Best accounts for practical AI engineering and tooling

Andrej Karpathy (@karpathy), Harrison Chase (@hwchase17), and Jerry Liu (@jerryjliu0)

These accounts are worth following if you care about turning model ideas into working systems, agents, and production tooling.

Andrej Karpathy stands out for showing how model intuition becomes code. Projects like nanoGPT and llm.c remove layers of abstraction, so it’s easier to inspect how training works under the hood. In March 2026, he released autoresearch, an agentic system that runs nanochat model training experiments on a single consumer GPU. The point is simple but powerful: it closes the loop between hypothesis, training, and results without human intervention. He also coined the term vibe coding to describe AI-assisted development where engineers focus on high-level orchestration instead of writing every line by hand.

Harrison Chase, the founder of LangChain, focuses on agent patterns that show up in day-to-day work: memory management, tool routing, retrieval chains, and multi-step workflows. A big reason people follow him is that he often replies directly to developers who are trying to fix actual problems.

Jerry Liu is the name to watch for LlamaIndex and the messy space between a nice demo and a system you can ship. His posts go deep on RAG orchestration and indexing strategies, and evaluation practices - the parts that start to matter the moment a notebook turns into production. If you build RAG apps, his feed is about as useful as it gets.

John Carmack (@ID_AA_Carmack) and key infra, MLOps, and evaluation engineers

Once you move past agent patterns, the next layer is deployment, reliability, and measurement.

John Carmack brings a systems-first point of view. He tends to push AI discussion back toward measurable tradeoffs, which is often what teams need when hype starts to drown out engineering judgment.

For production reliability, Vin Vashishta writes about deployment reliability - why deployments fail, how teams are structured, and what reliable deployment takes. Simon Willison is a go-to source for LLM security, especially prompt injection, along with practical CLI tooling and local model execution. Aman Sanger shares breakdowns of Cursor's internal routing and the telemetry behind how developers build with AI.

Account	Primary focus	Best for
@karpathy	Fundamentals & training	Minimal codebases, training from scratch, agentic systems
@hwchase17	Frameworks & agents	LangChain patterns, memory, tool use
@jerryjliu0	RAG & indexing	Retrieval quality, evaluation, production RAG
@ID_AA_Carmack	Systems & performance	Low-level optimization, concrete tradeoffs
@vinvashishta	MLOps & reliability	Production failures, deployment strategy
@simonw	Security & tooling	Prompt injection, local model execution
@amanrsanger	IDE & telemetry	Cursor internals, model routing

Research-grounded voices and a quick comparison table

Yann LeCun (@ylecun), Lex Fridman (@lexfridman), and Kai-Fu Lee (@kaifulee)

After tooling and infra, these accounts add research and market context that helps keep your feed grounded.

Follow Yann LeCun for a skeptical, research-first take on LLM limits, world models, and AI architecture.

Follow Lex Fridman for long-form research context turned into short X threads and podcast takeaways.

Follow Kai-Fu Lee for AI adoption and market signals outside the U.S., especially across Asia.

Comparison table: who to follow for news, tooling, infra, and research

Use the table below to find the account that matches the type of signal you want most.

Account	Best For	Why Follow
Sam Altman (@sama)	Lab direction, AGI timelines	First signal on OpenAI direction and releases
Andrej Karpathy (@karpathy)	Model training, AI education	Deep technical threads, minimal codebases, agentic systems
Yann LeCun (@ylecun)	Architecture debates, LLM skepticism	Research-first pushback on scaling narratives
Andrew Ng (@AndrewYNg)	Practical AI, industry adoption	Easy-to-follow implementation and agentic AI guidance
Dario Amodei (@darioamodei)	Scaling laws, safety engineering	Infrequent but high-signal essays on alignment and scaling
Harrison Chase (@hwchase17)	Agentic frameworks, orchestration	LangChain patterns, memory, tool use, direct replies
Jerry Liu (@jerryjliu0)	RAG, production data pipelines	Retrieval quality, indexing strategy, evaluation
John Carmack (@ID_AA_Carmack)	Systems, infrastructure	Low-level optimization, concrete engineering tradeoffs
Lex Fridman (@lexfridman)	Long-form research context	Connects researcher interviews to useful takeaways
Kai-Fu Lee (@kaifulee)	Global AI markets	Non-U.S. market signals, especially across Asia

Conclusion: Build a focused AI signal feed with X and daily.dev

X is still the fastest place to spot AI shifts in the moment. One post from the right engineer can take off fast and change how you look at a model, a tool, or a deployment call. The accounts in this guide give you a strong starting point if you want better signal on models, tooling, systems, and production tradeoffs. After that, the job is simple: turn that stream into a feed you can actually use.

The default timeline is built for engagement, not engineering depth. So be ruthless with curation. Set up a private "AI Core" list with 10–15 high-signal accounts from this guide. Group them around the three buckets covered above: news, tooling, and research. Then switch on push alerts for only the 3–4 accounts that have a direct effect on your production infrastructure. That way, you keep the signal and cut the noise.

For the context behind those posts, daily.dev adds depth. It brings in the longer write-ups and threads behind what you see on X, so you can understand the thinking behind the trend instead of just skimming the headline.

Use X for speed, daily.dev for depth.

FAQs

Which X accounts are best for LLM news?

For LLM news on X, the strongest signals usually come from researchers, practitioners, and lab leaders who share model releases, research updates, and architecture insights in real time.

A few accounts are worth watching. @_akhaliq is great for papers and launches. @sama, @demishassabis, and @DarioAmodei often post major lab updates. And @simonw, @karpathy, and @swyx are go-to follows for practical tooling and hands-on experiments.

If you want an easier way to keep up, daily.dev can help surface their best posts and threads.

Who should I follow for practical AI engineering?

For practical AI engineering, it helps to follow people who show how the work gets done: real workflows, agent-style coding patterns, and what production rollouts look like in practice. Industry news has its place, but it won't teach you much on its own.

daily.dev is a good place to find some of their best threads and longer write-ups without having to dig through everything yourself.

A few names worth following:

• Matt Pocock
• Boris Cherny
• Logan Kilpatrick
• Michael Truell and Aman Sanger
• Jason Wu
• Greg Brockman

How many AI accounts should I put in my X list?

Start with a small, high-signal list of 3–5 AI accounts. That lines up with the “pick 3–5 voices” guidance, helps sharpen your X feed, and makes it less likely that practical, substantive posts get buried in the noise.

If your focus still feels strong after a week, add more bit by bit.