Brian Burns
@brian_a_burns
Followers
2,895
Following
1,469
Media
223
Statuses
2,700
Recruiter & employee #12 at @harvey__ai . UW ML PhD dropout, started @ai__pub , ran an ML recruiting business.
San Francisco, CA
Joined October 2021
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
THE LOYAL PIN FRENCH KISS
• 1051095 Tweets
نابولي
• 189251 Tweets
Endrick
• 189241 Tweets
Mbappe
• 156622 Tweets
Fred Fred Fred
• 109562 Tweets
#AEWAllIn
• 105301 Tweets
ولي العهد
• 94871 Tweets
Oasis
• 57103 Tweets
#ولي_العهد_في_EsportsWorldCup
• 55732 Tweets
Mikel
• 46935 Tweets
Gökhan Akkan
• 43849 Tweets
Rize
• 34983 Tweets
O Corinthians
• 32629 Tweets
Independiente
• 32537 Tweets
Nigel
• 30328 Tweets
Mariah
• 29102 Tweets
Soto
• 28339 Tweets
Hakem
• 27181 Tweets
Ebuka
• 25650 Tweets
Fortaleza
• 24818 Tweets
Bryan Danielson
• 23683 Tweets
Pedro Raul
• 21211 Tweets
#OMSDR
• 19616 Tweets
#يلو_ترعي_o15
• 19561 Tweets
Kaleci
• 19408 Tweets
#ÇRSvFB
• 19335 Tweets
Dzeko
• 18744 Tweets
Swerve
• 18489 Tweets
Aaron Judge
• 18268 Tweets
Wahi
• 18156 Tweets
Turgut Doman
• 15379 Tweets
Ospreay
• 13709 Tweets
Reims
• 12968 Tweets
Muslera
• 12926 Tweets
Girona
• 11857 Tweets
Britt
• 10730 Tweets
Stanton
• 10381 Tweets
Unexpected date spot last night - the Berkeley City Club is a hidden gem
Very relaxed bar + lounge upstairs. Will be back!
21
40
2K
@debarghya_das
Next-level boggling to me to log on LinkedIn and see Google engineers publicly writing negatively and in long form about their employer.
(Saw another one of these yesterday!)
2
5
244
I started going to the gym when I was a 15 year old math nerd, mostly to put on muscle to pick up girls.
I didn’t like it at the start, so I decided to Pavlovian train myself by smiling every time I lifted a weight.
9
4
234
@proetrie
Have gone on a few first dates lately where the women insisted multiple times I go to therapy.
Maybe I'm psycho... but it's extremely offputting! Don't tell me how to live my life dude, it's a first date
11
0
223
Years ago I got an emergent ventures grant to map all mentorship relations between elite mathematicians since 1750
Finished the project but really should publish something… very interesting findings
8
24
233
@KyleACooke
Bane scene but for this plane
“So what’s the next step of your master plan?”
“Crashing this plane… with no survivors”
1
5
198
@tszzl
One of the best things I did going to a high school like that was read Cal Newport’s book on alternate college admissions strategies - basically drop all the AP classes/debate clubs and get good at weird things you care about.
More fun and easier way to get into Harvard/MIT etc
5
8
199
// Personal announcement //
I've closed down AI Pub's recruiting business to lead talent and recruiting at Harvey! ()
I joined ~2 weeks ago as employee
#12
.
1/9
21
5
179
We’re hiring across our research team at Harvey!
We have insane product market fit (0 to $15M ARR in 12 months)
And are tackling very hard research problems (LLM agents to do complex legal work like billion-dollar corporate mergers)
Apply at or DM me!
New Research Roles
We're scaling our research team at Harvey.
- Build foundation models, agents, search and eval systems to do the world's most complex legal work
- Traction: $0-15M ARR in the last 12 months
- $106M raised from OpenAI, Sequoia, KP
New roles below:
5
17
74
6
7
166
Over the last month I grew
@ai__pub
from 0 to 10.4k followers on Twitter.
Huge surprise for me - but I was also very deliberate & principled about growing the account.
Some friends have asked, so: my opinionated thoughts on how Twitter works and how to grow an account.
1/22
6
9
158
Harvey is the fastest-growing LLM startup I'm aware of - they just announced a $21M Series A led by Sequoia.
They're working on insanely hard engineering problems and are building a world-class team.
I recruit for them - apply here:
Harvey just announced a $21M Series A led by Sequoia.
Backed by the OpenAI Startup Fund since August 2022, Harvey is building AI knowledge workers for law.
They're the fastest-growing LLM startup I know of, and are hiring aggressively (below).
1/5
4
13
99
2
11
153
@Molson_Hart
IIRC Charlie Munger commented that this was one of the major hidden benefits of Costco membership for middle-class Americans
5
0
143
Best way to beat the “Sunday scaries” is to just work all day on Sunday
4
2
118
We just raised $100M at $1.5B and are hiring:
- Research Scientists, Research Engineers, and AI Infrastructure Engineers
- Product Engineers and Infrastructure Engineers
all at the Staff+ level - come work with us!
Feel free to dm me or apply here:
@harvey__ai
raised a $100M Series C from
@GVteam
,
@OpenAI
,
@sequoia
,
@kleinerperkins
,
@eladgil
, and
@svangel
at a $1.5B valuation. We’re honored to work with such an amazing team, clients, and investors.
13
29
146
5
7
116
@kitten_beloved
Stay-at-home-girlfriend without marriage sounds like such a precarious position
2
0
107
@tszzl
(To fill in details, I took the minimum load of classes, taught myself a ton of math, almost went to prison for animal rights activism (breaking into factory farms & rescuing animals), got into every college I applied to)
5
0
99
Harvey is partnering with PwC and OpenAI to build domain-specific foundation models for tax, legal, and HR services.
Announcement:
4
7
100
Not sure if it worked to motivate me, but the conditioning persists!
11 years later, massive grin for each rep.
😁🏋🏼♂️
2
1
93
At a company offsite with ~70 other people in Austin right now
A year ago Harvey had precisely 2 employees :)
7
1
95
@FritzTheDev
@paulg
AI safety is real IMO, almost every instance I’ve seen of “AI ethics” is political nonsense
2
1
87
Harvey has raised an $80M Series B led by
@eladgil
and
@kleinerperkins
, with participation from the OpenAI Startup Fund and
@sequoia
.
19
23
235
6
0
79
@debarghya_das
PG has a good essay that touches on this topic at the beginning:
“So I started to pay attention to how fortunes are lost. If you'd asked me as a kid how rich people became poor, I'd have said by spending all their money…
3
4
70
@JosephJacks_
I recruit for a bunch of LLM startups and haven’t seen anything close to this. Best candidates (engineering leaders, eg principal or VP level) getting $1M max
What’s the highest offer you’ve actually seen? Sounds like hearsay tbh
7
0
67
Finally earned it!
(No idea why anyone takes this guy seriously anymore)
15
0
63
Ignoring doctors’ medical advice in favor of that from anonymous cartoon characters on Twitter remains one of life’s greatest hacks
3
6
64
Grothendieck, on "the capacity to be alone":
"In those critical years I learned how to be alone. [But even] this formulation doesn’t really capture my meaning. I didn’t, in any literal sense, learn to be alone, for the simple reason that this knowledge had never been unlearned
0
10
57
@growing_daniel
Personally have the opposite preference.
I'd much rather pair early and grow as people together than pair after I have (and she has) "figured everything out".
Makes the bond much deeper and more meaningful.
4
0
54
@callicrates_
@rivatez
would tell people at parties that she was a professional lizard breeder
3
1
48
First
@ai__pub
meetup in SF was off the wall!
100 people showed up, 6 XL pizzas consumed, many startup connections made, piñata demolished
3
0
46
Non-tier-1 VCs like to pretend they are capital allocators and come up with “investment theses”,
as if they had a huge plate of excellent deals to choose from - among which they must intelligently allocate capital.
Nope!
Non-tier-1 VCs are fractional service providers.
10
3
47
The market for talent at early-stage startups (~Series A) is one of the most illiquid and inefficient markets I know of
5
0
47
At Harvey we evaluate LLM systems by having in-house lawyers from top-5 BigLaw firms work with them live, on complex legal tasks they are familiar with from years of experience.
(In addition to LLM eval through benchmarks, and LLM systems for LLM eval)
imagine if a job interview was looking at the grades someone got without talking to them or working with them whatsoever
this is how llm evals work
8
24
342
1
3
38
I turn 28 today! (On March 28).
Am proud to share a birthday with my childhood hero Alexander Grothendieck, arguably the greatest mathematician of the 20th century.
He grew up in refugee camps and wrote extensively on personal and spiritual matters - a quote of his below.
Grothendieck, on "the capacity to be alone":
"In those critical years I learned how to be alone. [But even] this formulation doesn’t really capture my meaning. I didn’t, in any literal sense, learn to be alone, for the simple reason that this knowledge had never been unlearned
0
10
57
4
0
39
~$50k a placement in tech for those doing solo contingent recruiting.
Underrated "starter business" for those wanting to try small business entrepreneurship: low startup costs, fragmented market, not much competition, etc.
(This is what I did after taking leave from an ML PhD).
Want an idea of how lucrative recruiting can be?
A "tough year" is making $600k 🤯
7
1
30
3
0
39
1) Twitter typology.
All large accounts provide some "service" to their followers, in *at least* one of these buckets:
a) Make you smarter
b) Make you money
c) Provide entertainment
d) Provide parasocial relationship
2/22
1
0
35
@ethanCaballero
From what I've seen / heard it won't be multimodal, if you read the article they constructed the headline from two different bits of information:
1) GPT-4 coming soon
2) New multimodal models coming soon.
As usual, shaking my head at journalists...
1
0
35
Have started to use little combinatorics problems for work breaks
2
1
35
@thomasschulzz
Counterpoints:
- Elon started Zip2 before PayPal, Tesla, SpaceX etc
- Bezos worked in quant funds until he was ~30 before starting Amazon
- Paul Graham explicitly started Viaweb to get rich, did YC later (in his late 30s/early 40s?) for impact
Pretty tried and true model IMO
5
1
34
Some problems in life are incremental:
Eat better, exercise a bit more, and you’ll lose some weight.
Other problems in life are like building a car:
You have to get *everything* right, otherwise you fail. Conversely - when you get all the components right, you win big.
4
0
32
Having to negotiate more frequently in professional life, so wanted to pick up the basics.
Recommended by a polymath friend - it’s great so far!
3
1
32
My weird intro to programming was learning Mathematica to solve combinatorics and Project Euler problems in high school.
Clicked on an ad in the Wolfram Alpha sidebar one day, got invited to a free summer camp - became a better programmer and met Stephen Wolfram!
3
0
32
@aniiyengar
FWIW there are also interviews of him going back to high school where he talks about needing to become very wealthy in order to build space colonies.
Life's work = space exploration, Amazon was him figuring out how to get rich to get there.
2
0
32
@finbarrtimbers
Short version: S-tier engineers would love to know who the great seed/series A companies are so they can dip from Google and join, and founders would love to have them!
But founders are typically too busy at this stage to be marketing themselves to devs or doing *serious* cold
3
1
31
@FusorFusion
Lol, it turned out being lanky was not the main barrier to getting girls for me. But it helped!
1
1
30
@BowTiedGatorDDS
@BowTiedOx
@Grimhood
@callicrates_
My industry friend recommended I try your hydroxyapatite toothpaste!
He said he went to a dentist, had two cavities, ignored the dentist, used your toothpaste for a couple months, came back, now has one cavity
1
1
30
Have seen Lindy firsthand - it’s a wild product! On-demand teams of usable AI agents will permanently change the economy.
1/ Announcing the new Lindy: the first platform letting you build a team of AI employees that work together to perform any task — 100x better, faster and cheaper than humans would.
Watch how I build a Lindy to track competitors in 30s:
95
212
2K
1
2
30
@robinhanson
Such that, if you're going to kill one person in the category, it makes sense to kill all of them.
Therefore honest answers to this question are either:
0: opposed to killing
<5: personal vengeance
>100: you've found a category
Not much in between.
3
0
29
Harvey featured in Bloomberg Law today:
1
0
28
Outside of the safety teams at actual AI orgs like OpenAI, Anthropic etc,
This is the embarrassing sense I’ve gotten speaking with every effective altruist or rationalist “working” on AI Safety.
No specific plan for what to do, no specific sense of what’s going on.
Ezra Klein with a critique of Effective Altruists' theory of change in AI policy
20
65
698
5
3
29
I was able to get to 10k by only tweeting 2-3x/week.
But these tweets were *very high value*, long-form threads that took several hours to research and write.
This thread, e.g., took me a full day to research and put together, but lifted AI Pub from 100 to 3k followers.
8/22
// Stable Diffusion, Explained //
You've seen the Stable Diffusion AI art all over Twitter.
But how does Stable Diffusion _work_?
A thread explaining diffusion models, latent space representations, and context injection:
1/15
89
1K
5K
1
0
26
The Supreme Court of Singapore is partnering with Harvey to increase access to justice:
0
4
25
@robinhanson
Pretty screwed up thought, but:
Outside of personal vengeance (where the TAM is <5 for most people), any other kill target comes in a "category" with >100 people in it.
E.g. geopolitically relevant and maligned actors, criminal kingpins selling harmful drugs...
1
0
24
Calm down everyone, he’s just taking the first term of the Taylor series!
3
1
24
Join Harvey
1
0
26
@Pavel_Asparagus
Also - once you find your purpose, you need instrumental skills and resources to realize it: a nuanced worldview / a secret, a network, a team, access to capital, specialized skills (e.g. engineering) etc.
You can build the latter up on the road to discovering the former.
0
1
23
@sashachapin
I put up with this for years, now happily divorced. Slack and forgiveness (frankly, even kindness) are truly foreign and an incredible breath of fresh air now
1
0
23
EA deserves disrepute since it is increasingly becoming a combination of The Gates Foundation and the Democratic Party,
Organizations that EA originally positioned itself against as bloated political or pseudo-political organizations beset by grifters.
2
0
21
Posting this because my feed is filled with tier 2 and below VCs posting their “theses” on AI (which are at least a year behind the curve) while none of them have means of adding serious value of a portfolio company
And therefore will win no deals
And clog the feed with garbage
4
0
22
@GraduatedBen
35% - oddly close to the fraction of new dollars printed last year. Funny how that works - a bunch of NYT economists told me otherwise…
1
0
21
Sending a cold email is an everyday example of taking asymmetric bets
6
0
23
@tszzl
Maybe dumb but is the modern equivalent of that claiming that we’ll reach “AGI” by just scaling up params/data of SSL transformer models?
0
0
20
@callicrates_
Paul Graham has an interesting arc here. Was a broke painter/hacker until his early 30s, when he realized he should get rich - just for the sake of being set so he could paint and hack.
Started a startup with the express purpose of making a few million and sold immediately…
1
1
22
Since, the community has:
- generated essentially no new ideas,
- seen a brain drain of talent, and
- adopted an increased clamoring for philanthropy-grifting and “regulation”.
2
1
20
// Harvey: New Roles //
We're ramping up hiring!
If you want to work on unprecedented legal AI, apply here:
Now hiring (Senior / Staff):
- Backend, Frontend, Product Engineers & Manager
- ML Research Engineers, Scientists & Manager
- Security Engineer
6
1
23
@tszzl
Spoke with Dan Fu and Tri Dao at Stanford, the authors of the H3 paper - gives n*log(n) scaling for context length instead of n^2 for attention, with similar/same performance as Transformers.
They mentioned all of this stuff (state space models) has been around since the 60s
1
0
21
It was game over for EA in 2016 when EA leadership decided that cold-calling for the Hillary campaign was an “effective cause area”.
1
0
19
@GSVBemusementPk
Yeah same here. Main motivation for me is having consistently higher energy/focus - used to have bad sleep issues.
0
0
21
@noahkreutter
@astridwilde1
Huge diff too between AI Safety (real serious concerns) and “AI Ethics” (woke political bullshit) - the outcry about open source falls squarely into latter camp
1
0
22
I tend to play with my cards close to my chest, but:
2023 is going to be an incredible year for AI Pub.
Will be the best Twitter-native technical AI publication for AI developers and researchers.
Currently taking steps to grow it beyond a one-man show.
1
2
22
@KTmBoyle
If you think this is bad, should know what happened with Hikvision a few years ago - Chinese surveillance cameras with known vulnerabilities and back doors installed on hundreds of US military bases
1
2
20
// Harvey: Frontend Engineer //
Harvey is looking to hire our first front-end engineer.
We're building LLMs and LLM systems for legal, are backed by Sequoia and the OpenAI Startup Fund, and by all metrics have incredible traction (revenue growth, headcount growth).
1/3
2
3
21
