Percy Liang✓ Add us as a preferred source on Google
🔄 Data last refreshed 1 month ago
🤝
Make a deal with Percy Liang
🔒 Paid proposals held safely in escrow — released only when the work's approved.
Followers
112.8K
Posts
1.3K
Following
425
Account age
16 yrs
💰 Suggested sponsorship rate
$632 – $1K / sponsored post
Estimated from audience size & typical reach — every deal is negotiable and held in SocialDB escrow.
🎁 Free analysisfor Percy Liang
📊 Post performance
Recent posts · views
Dec 15latest
226.7K
Avg views / post
104.1K–425.6K
Typical post (views)
2×
Reach vs followers
66/100
Consistent reach
Oct 2009
On X since
Half of recent posts land between 104.1K and 425.6K views (median 261K) — a dependable floor a sponsored post can count on.
🔥 Top post: While we celebrate @deepseek_ai 's release of open-weight models · 776.5K views
📊 Activity & format
Posting cadence
0.1 / week
A lower-frequency account — each post lands with more weight.
Content mix
Mostly text
Recent: 1 video · 1 image · 16 text.
Follower / following
265×
Follows 425 back. A strong ratio — an audience that follows them, not a follow-for-follow network.
Verification
✓ X Premium
A paid X Premium (Blue) verified account.
Recent posts
View on X ↗
You spend $1B training a model A.
Someone on your team leaves and launches their own model API B.
You're suspicious. Was B was derived (e.g., fine-tuned) from A?
But you only have blackbox access to B...
With our paper, you can still tell …
-2016 (classic era): focus on data efficiency
2017-2025 (pretraining era): focus on compute efficiency
2026-: focus on data efficiency (again)
The standard Transformer paradigm is optimized for compute efficiency. As we look at data effici…
We did a very careful study of 10 optimizers with no horse in the race. Despite all the excitement about Muon, Mars, Kron, Soap, etc., at the end of the day, if you tune the hyperparameters rigorously and scale up, the speedup over AdamW …
Wrapped up Stanford CS336 (Language Models from Scratch), taught with an amazing team @tatsu_hashimoto @marcelroed @neilbband @rckpudi. Researchers are becoming detached from the technical details of how LMs work. In CS336, we try to fix t…
What would truly open-source AI look like? Not just open weights, open code/data, but *open development*, where the entire research and development process is public *and* anyone can contribute. We built Marin, an open lab, to fulfill this…
So why are DeepSeek models so good? To their credit, the papers have more detail than most frontier model papers, but it’s hard to tell which details matter. And data is the big missing piece, which we know to be the most important factor …
True open-source allows us to study and modify artifacts. We can study the DeepSeek papers (which are nicely written, but still omit details), and we can fine-tune their models. but one cannot understand or modify them at a deep level.
I miss the days when we evaluated algorithms rather than models. Rather than "how well does model M do?", it should be "given data D and compute C, how well does running algorithm A on D with C do?" I don't think we can get scientific c…
This year, I have 4 exceptional students on the academic job market, and they couldn’t be more diffferent, with research spanning AI policy, robotics, NLP, and HCI. Here’s a brief summary of their research, along with one representative wo…
We should call models like Llama 3, Mixtral, etc. “open-weight models”, not “open-source models”. For a model to be open-source, the code and training data need to be public (good examples: GPT-J, OLMo, RedPajama, StarCoder, K2, etc.). Wei…
Myth: open foundation models are antithetical to AI safety.
Fact: open foundation models are critical for AI safety.
Here are three reasons why:
What would truly open-source AI look like? Not just open weights, open code/data, but *open development*, where the entire research and development process is public *and* anyone can contribute. We built Marin, an open lab, to fulfill this…
So why are DeepSeek models so good? To their credit, the papers have more detail than most frontier model papers, but it’s hard to tell which details matter. And data is the big missing piece, which we know to be the most important factor …
True open-source allows us to study and modify artifacts. We can study the DeepSeek papers (which are nicely written, but still omit details), and we can fine-tune their models. but one cannot understand or modify them at a deep level.
I miss the days when we evaluated algorithms rather than models. Rather than "how well does model M do?", it should be "given data D and compute C, how well does running algorithm A on D with C do?" I don't think we can get scientific c…
This year, I have 4 exceptional students on the academic job market, and they couldn’t be more diffferent, with research spanning AI policy, robotics, NLP, and HCI. Here’s a brief summary of their research, along with one representative wo…
We should call models like Llama 3, Mixtral, etc. “open-weight models”, not “open-source models”. For a model to be open-source, the code and training data need to be public (good examples: GPT-J, OLMo, RedPajama, StarCoder, K2, etc.). Wei…
Myth: open foundation models are antithetical to AI safety.
Fact: open foundation models are critical for AI safety.
Here are three reasons why:
💡 Facts
🗓️Joined X in 2009 — 16 years ago.
👁️Averages 226.7K views per post.
📤Posts about 0.1× per week.
💬1.46% engagement rate.
🏅Earned the 100K Followers for passing 100K followers.
🕵️ Fake follower check
Estimated83/100
Excellent
Credibility score
93%
Real
Real audience
Low
Fake-follower risk
High
Data confidence
- Est. 93% real, active audience · Low fake-follower risk.
- Strong engagement (~2× of followers engage each post) — an active, real audience.
- Organic base — far more followers than accounts it follows.
- Verified account.
- Established account (16+ years old).
- 1.46% engagement — below the ~14.4% typical for this size.
Heuristic estimate from engagement, follower ratios, account age & growth — a screening signal, not a guarantee.
About
professor of computer science @Stanford @stanfordnlp, co-founder of @togethercompute, http://marin.community, @simile_ai, pianist
🔀 Audience overlap
EstimatedEstimated shared audience with similar creators — useful for avoiding overlap (or doubling down) when planning a campaign.
More like this
Find more →
Frequently asked questions
How much does Percy Liang charge for a sponsorship?
Percy Liang hasn't published fixed prices yet. Send a proposal through SocialDB and agree terms directly — your payment is held in escrow until the work is approved.
How do I contact Percy Liang for a brand deal or collaboration?
Send a paid deal or collab request right here on SocialDB — we notify Percy Liang, and once they accept, your payment is held safely in escrow and released when the work is approved. You don't need to track down an email.
How many followers does Percy Liang have?
Percy Liang has 112,820 followers on X.
✉ Message Percy Liang
Reaching out to influencers is a Pro feature. Upgrade to message any influencer directly — perfect for brands and agencies booking sponsorships.
- ✓ Message any influencer from their listing
- ✓ The influencer gets notified by email
- ✓ Manage every conversation in one inbox
Already Pro? Log in.
🎤 Event / appearance with Percy Liang
Booking an event / appearance is a Pro feature. Upgrade to book Percy Liang for an in-person or virtual appearance — payment held safely in escrow until the event is done.
- ✓ Book them for events, livestreams, panels & more
- ✓ Percy Liang gets notified by email
- ✓ Fee held in escrow, released after the appearance
Already Pro? Log in.
🚀
You're out of free requests this month
Free accounts get 5 per month. Go Pro for unlimited sponsor pitches, collab requests & sponsorship deals — plus featured placement, the Verified badge, free withdrawals and more.
Upgrade to Pro — $9.95/mo →Your free limit resets on the 1st of next month.
Auto-generated from X's public data — no affiliation with or endorsement by SocialDB. Is this you? Claim it · Remove