Hugo Bowne-Anderson
@hugobowne
Followers
14,395
Following
3,585
Media
669
Statuses
10,920
data scientist, writer, educator, anecdotalist, host of @vanishingdata podcast
Jupyter
Joined January 2013
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
ドーナツホール
• 192737 Tweets
#今月描いた絵を晒そう
• 160651 Tweets
Mombasa
• 51178 Tweets
Griezmann
• 44566 Tweets
#INDvBAN
• 33608 Tweets
#JusticeForBruce
• 27373 Tweets
#CDTVライブライブ
• 25814 Tweets
#中島健人のヒトゴト
• 17016 Tweets
スタリラ
• 16994 Tweets
おはスタ
• 15213 Tweets
Grizou
• 15097 Tweets
نعيم قاسم
• 13930 Tweets
Deschamps
• 10908 Tweets
Pinned Tweet
💫 I recently did a 3 hour livestream w/
@eugeneyan
,
@BEBischof
,
@charles_irl
,
@HamelHusain
, and
@sh_reya
(we missed
@jxnlco
!) about their recent report on 42 Lessons Learned from a Year of Building with LLMs.
We've just now released two
@VanishingData
podcast episodes of this
1
4
42
A former colleague of mine told me of a company where, in meetings, everybody began with 3 cards. Each time you spoke you had to put down a card. Once all 3 were down, you had to wait until everyone had put down theirs.
Men in the meetings resorted to trying to buy other’s cards.
17
198
784
👋
@karpathy
: we're organizing an LLM course that has exploded into a conference. We'd love to have you present, if you're available. Totally np if not of c but I'd regret not asking you. lmk! 💫
This is our current lineup:
22
44
653
I stand with Kara Woo and I support the Rstats & broader data science community in taking a strong stance on this critical issue. I am deeply saddened that the community has not yet been sufficiently listened to by DataCamp. 1/
23
51
523
I recently had my last day at
@DataCamp
. I feel privileged and grateful for the opportunity I have had to reach so many people motivated to continuously learn, who recognise how important data literacy and data skills are, while interrogating their uses. 1/
73
11
456
I am so stoked to have recently joined
@CoiledHQ
as Head of Data Science Evangelism & Marketing. And I'm super excited to be bringing you all a new world of data science content & to build it with you. Why did I join Coiled? Read on, my friend:
37
17
279
If you wanna do data science, learn how it is a technical, cultural, economic, and social discipline that has the ability to consolidate and re-arrange societal power structures.
8
60
278
What are the biggest challenges facing data science, ML, and AI as an industry?
89
43
238
wtf is this sorcery
7
30
217
OK, let's start getting real about LLMs again. Join me and
@jeremyphoward
later this week as we discuss the current and future states of generative AI.
We'll talk about
* What on earth we’ve just witnessed at OpenAI, why it’s important (and why it’s
5
36
215
Startups have shamelessly re-branded rudimentary machine-learning algorithms as the dawn of the singularity, aided by investors & analysts who've a vested interest in building up the hype. Welcome to the artificial intelligence bullshit-industrial complex.
4
64
211
If you wanna do data science, learn to communicate effectively with non-technical stakeholders.
9
30
206
To all you wonderful
#DataFramed
listeners, I'm sincerely sorry that I've stopped releasing episodes w/o notice. Know that I appreciate you deeply & that I also appreciate your patience during this challenging time. I plan to bring back the DataFramed podcast in the near future.
11
4
173
Do you know of any real-world datasets that can be used to demonstrate common distributions? E.g.
* Poisson
* Exponential
* Gaussian
* Negative binomial
* Gamma
* Beta
And more! This list goes on!
28
25
170
I wrote some words for
@HarvardBiz
, synthesizing what I've learned from my conversations with 35+ data scientists on
#DataFramed
, the
@DataCamp
podcast:
2
44
167
Who else has been working in PythonLand so long that the planet (and god) name “Jupiter” consistently looks misspelled to them?
9
5
164
I'm excited to be teaching Deep Learning from Scratch w/ PyTorch at
@SciPyConf
in ~40 min w/
@dhavidearuliah
. Here's the material:
and here's where SciPy attendees are joining from!
5
27
161
If you wanna do data science, learn regular expressions.
12
13
146
I taught a
#PyCon2019
tutorial w/ my friend
@ericmjl
on
#Bayesian
inference, probability, data-generation storytelling, simulation & probabilistic programming. Watch the video here:
1
33
145
And we're live! We're super proud to share the 1st 6 episodes of
#DataFramed
, a
@DataCamp
podcast. Subscribe on iTunes/Playstore/app of choice. Weekly from here on in.
11
43
134
If you wanna do data science, learn about cognitive biases, our alarming lack of statistical intuition, and how to correct for them.
6
17
132
Let’s see Jonathan Cornelissen open that door. I hope to play a very active role in helping DataCamp rebuild that trust. 3/3
6
4
134
High level points
@jxnlco
and I covered today in our "How to Build Terrible AI Systems" livestream for
@VanishingData
.
Watch here
More details in 🧵 thanks to
@AnthropicAI
& Claude Opus
1/
1
19
132
I'm back from a month off twitter. It was one of the best months I've had in a long time. As we know, correlation doesn't imply causation but it sure is correlated with it. Oh, and also:
#DataFramed
, the
@DataCamp
podcast, is coming back atcha next week :)
Happy holidays, EVERYBODY! I for one am going on a social media detox, which means (among other things), I won't be on twitter much for the next month. Wish me luck! 🤓
0
0
14
9
4
127
I am super proud to be launching DataFramed next week, a
@DataCamp
podcast about all things
#datascience
. Watch this space.
We’re launching a new podcast called DataFramed! Check out this article with a sneak peek trailer to see what to expect from this new podcast by
@hugobowne
set to launch next week!
4
41
102
5
32
125
✨I'm way too excited to be having a fireside chat with my old pal
@jeremyphoward
on LLMs for Hackers. J is a true inspiration to me here so can't wait to grill him! 📺
Sign up for free below:
3
15
126
New DataFramed podcast recording studio in Sydney for
@DataCamp
. We're pretty darn excited about the interviews we have coming up. You should be too.
7
8
122
Along with other employees, I expect to see more accountability & transparency from our CEO in the coming days & from our company as a whole going forward. It will take a long time to regain the trust of the community. That needs to start now. 2/
3
4
119
I am really excited that
@_inesmontani
's
@DataCamp
course on
#NLProc
w/
@spacy_io
is out. I've been wanting us to build this course for years.
5
20
119
man
@jxnlco
's posts are great. check this one on How to Build a Terrible RAG System
The Ingredients for Disaster
- Don’t Worry About Latency
- Never Show Intermediate Results
- Ignore User-Specific Queries
- Skip Fine-Tuning
1
15
113
man going live with
@sh_reya
in just over 30 min and devouring this. not sure how she publishes so much high quality work
4
11
111
Do you have any data quality, data cleaning or data preparation stories that you can share with me? (horror stories or otherwise). I'm writing an article on this. Feel free to comment and/or DM.
42
18
106
🎙️High level points we covered today in "Lessons from a Year of Building with LLMs" with
@eugeneyan
@BEBischof
@sh_reya
@HamelHusain
@charles_irl
@jxnlco
Watch here:
@vanishingdata
podcast out next week 💫
1
28
108
In 2019, I (re-)read more non-fiction than fiction (rare for me), much of it related to the impact of the planetary-scale inference-and-decision-making systems we're building that are conceptually &/or practically flawed, to paraphrase Michael I. Jordan. Serious highlights: 1/
4
8
107
We're incredibly proud to be launching our 1st
#Dask
@DataCamp
@anacondainc
course : a flexible parallel computing library to easily scale
#Python
#datascience
workflows up to large datasets. Join inimitable
@dhavidearuliah
&
@mrocklin
for a great course!
3
42
105
People ask me all the time about getting started in data science. I say jump into data sets that interest you, discover things, put on GitHub, write a blog. Here’s some data that’ll interest many.
We published the largest database of university professors' conflicts of interest.
Let us know what you find....
(but actually, please do — databases like this are gold mines for new investigations, but even we can't review it all)
3
43
130
2
21
102
hey! I just gave a talk at
#PyDataNYC
on what I've discovered about
#datascience
while hosting
#DataFramed
, the
@DataCamp
podcast. Slides are here:
3
23
106
sometimes on friday evening i'll leave monday Hugo a cheeky code surprise such as `import pandas as np` just to mess with him.
3
8
104
🐍 Attention Python devs who have wrestled with dependency management! 🚨
@wuoulf
, the creator of Mamba 🔥, is back with Pixi 📦✨, the package manager that will make you forget your package management woes. 😌
In this live demo (full Outerbounds fireside chat in comment), Wolf
4
28
103
I am having a helluva time editing next week's ep. of
#DataFramed
, the
@DataCamp
podcast, a conversation w/
@wesmckinn
about
#datascience
tool building, past, present & future. Can't wait to drop the 2nd last episode of Season 1 🎙️
4
15
102
you need to play this game now:
4
47
101
I'm so stoked to be launching a new podcast. Here you can find a short trailer to set the scene 👇 In a few days, I'll be dropping a 2hr conversation with
@jeremyphoward
about making data science uncool again. Stay tuned! 🎙️
9
16
102
Such exciting news for
@DataCamp
! It has been a wild ride playing a small role in this company's growth from 10 to 100 employees -- and from teaching 300K learners to now 3million+ worldwide. We're hiring a lot so keep your eyes on our careers page
1
14
100
.
@mrocklin
& I have been chatting with lots of data scientists about the pain points of distributed computing. **We wrote up the patterns that emerged**. How do they resonate with you? Give us feedback! Disclaimer: we're building products to solve these.
8
23
99
If u like Regexes (and srsly, who doesn't), ur gonna ❤️ Regex Crosswords. Ur welcome.
5
22
97
This is a wonderful & intuitive post on the inspection paradox by
@AllenDowney
. Such clear, precise & elegant prose also! Looking forward to his talk on this at
#pydatanyc
0
38
97
STOP SAYING R CAN'T SCALE IN PRODUCTION (FOR ML OR OTHERWISE)
#rstatsnyc
#rstats
I left this fact out of my talk but at T-Mobile our R models are hit over 4 million times so next time someone tells you R can’t scale, @ me
#rstatsnyc
7
76
283
4
22
95
Keep your eyes on the docstrings.
Not that you're wrong.
10
4
95
Can you believe Ep. 40 of
#DataFramed
, the
@DataCamp
podcast, is coming out next week? Currently editing & it's a corker! Strap in for a convo w/
@BecomingDataSci
about becoming a data scientist... Renée rocks!
6
16
95
Nearly everybody: If you want to run machine learning in production you use python.
Me: You're wrong.
Nearly everybody: WTF, dude?!
Me: Check out ~18:45 of Ep 48 of
#DataFramed
, the
@DataCamp
, & listen to
@heatherklus
talk about using
#rstats
in production at T-Mobile. 🎙️🎤
Ho boy, I’m on this week’s
#DataFramed
segments!
I talk with
@hugobowne
about why data scientists should care about Docker and how we use
#rstats
in production at T-Mobile.
Give it a listen!
1
9
44
5
19
89
I've been thinking a lot about decision making recently & wrote an essay for
@radar
-- let's talk about uncertainty, probabilistic thinking, assessing risk, information quality, and how each and every report of statistics is a political act
5
12
91
why the constant language wars?
serious question.
doesn't seem constructive.
happy to be corrected.
R or Python for statistics? It's basically the same.
Example: fit a linear model
R:
m1 <- lm(y ~ x1 + x2)
Python:
import numpy as np
import statsmodels.api as sm
m1 = sm.OLS(y, sm.add_constant(np.array([x1, x2]).T))
85
308
2K
23
5
85
For the record: I’m totally serious.
I can't tell whether this guy doing the DataCamp video on Python is serious or not: "In this video, I'm going to introduce you to functions. Once you learn about them, you won't be able to stop using them. I sure can't!"
5
0
10
4
0
83
Whoa! Get ready for Ep. 26 of
#DataFramed
, the
@DataCamp
podcast, a chat w/ the inimitable
@JennyBryan
about
#spreadsheets
in
#datascience
. this one's a corker!
0
21
83
🎯
@HamelHusain
on AI development:
✅'Evals'? Forget the fancy term. It's about systematic improvement.
✅Stop eyeballing. Start measuring progress.
✅No structured approach? You're not really doing AI.
✅ It's not optional. It's how you build. Period.📊💡
(full convo w
3
10
83
Hard at work in the new office. Audrey's doing the heavy lifting. Best. Assistant. Ever.
3
0
83
Ep. 40 of
#DataFramed
, the
@DataCamp
podcast, is live!
@BecomingDataSci
& I chat paths to becoming a data scientist. We discuss common data scientist profiles, how to figure out which ones may be a fit for you & much more.
5
19
77
Day 3 of 14 in this hotel room guarded by police & military no windows open i'm not allowed to leave even for walks meals like this are delivered cold sometime in 2 hour windows and I now feel like we all including me have been using the term quarantine way too fucking liberally
14
4
77
I am proud to be launching Season 2 of
#DataFramed
, the
@DataCamp
, w/ a conversation w/
@csdoctorsister
about
#inclusivity
in
#DataScience
, barriers to entry for under-represented groups, initiatives that exist & what we can do to be as inclusive as possible.
6
15
79
I use both Python and R and in that order. I love
#excel
. I also hate excel. Find out reasons for love/hate relationships with excel in my convo w/
@JennyBryan
on
#DataFramed
, the
@DataCamp
podcast:
3
16
77
@PrasoonPratham
@marktenenholtz
This is good advice! I’d advise learning ML and Python in parallel (slight difference) as they inform each other.
You can also skip Python completely and work in
#RStats
Tidy models is great!
3
5
75
I'd like to podcast again at some point.
Who would you like me to interview and why?
18
1
77
🚀 I'm excited (and nervous!) to report that I'm going independent as an AI, data science, education, and community building strategy consultant. Drawing from my experiences at
@OuterboundsHQ
,
@CoiledHQ
,
@DataCamp
, and research at Yale University and the Max Planck Institute, I'm
9
4
78
Not only is next week's ep. of DataFramed, the
@DataCamp
pod, w/
@BecomingDataSci
, but coming soon are conversations I loved having with
@AllenDowney
,
@vboykis
,
@StatModeling
&
@ellisonbg
! I'll also soon be speaking w/
@quaesita
,
@kierisi
,
@bearloga
+ other special guests...
1
8
74
This is
@EWRairport
now. Only ghosts here. I’m trying to get to Sydney, where the Aussie govt will quarantine me for 14 days & enforce isolation, the right way to do this. I’m incredibly grateful for everybody working at the moment. Wish me luck getting there 🤞.
13
1
75
If you wanna do data science, it's good to know about Type III errors: the right answer... to the wrong question!
2
10
73
💫 Check out
@jxnlco
's LLM & RAG consulting playbook -- relevant to ML and AI work, more generally.
I feel like I should increase my hourly rate after ingesting it 🤑
Full conversation here:
1
7
74
What are your favourite resources on causality in data/stats/inference?
13
10
73
If you wanna do data science, learn the powerful art of simulation: turn your laptop into a lab! 💻🥼🧪
3
7
72
📢✨Interested in how to actually deploy LLMs to production and all the concerns and pitfalls?
I'll be livestreaming with my main man
@HamelHusain
to find out. You can sign up for free here: 🤓💻
0
12
70
You’d actually be amazed how many messages like this I get across all channels 😶
13
1
67
" 9 out of 10 people will tell you to build a chat bot with an LLM. YOUR JOB IS TO SAY NO!" --
@HamelHusain
from this course
5
5
66
I've been thinking a lot about what 21st century data literacy can look like & started writing it down for
@HarvardBiz
. Watch this space. Shout outs to
@AllenDowney
,
@mrogati
,
@_rchang
,
@FastForwardLabs
,
@FSchueuer
( many thx to
@hmason
for connect there)
4
17
65
Get ready for next week's ep. of
#DataFramed
, the
@DataCamp
podcast, a convo w/
@omojumiller
, snr
#machinelearning
data scientist
@github
, about the role of ds in product dev, how to “use computation to build products to solve real-life challenges” and much more. Omoju rocks!
2
13
62
i keep returning to this post by
@AllenDowney
& if anybody's listening, there is still only one test!
2
15
62
And professionally / creatively ? Watch this space. And never stop learning, asking questions, and listening! 4/4
6
0
63
which data science/ML packages do you think have good documentation and why?
31
3
62
@vboykis
Data science isn’t really a thing and if it is, y’all are rarely & barely scientific about any of it
4
5
62
I recently spoke about learning to code for traders, data science, education & more w/ deadset legend
@AaronFifield
on
@chatwithtraders
. For those lovers of
#DataFramed
, the
@DataCamp
podcast, who want to hear me on the other side of the interview, NOT TO BE MISSED.
2
10
62
How on earth does it simultaneously feel like there are way too many data tools and not enough data tools? 😿
10
1
61
we have ~20 open positions here
@DataCamp
; it has been a wild ride over the past couple of years and if you're interested in working with a fast-growing data science EdTech company, check out the openings here:
2
27
61
3 sleeps until liftoff for Season 2 of
#DataFramed
, the
@DataCamp
podcast. Join
@csdoctorsister
and me, as we discuss
#inclusivity
in
#DataScience
, barriers to entry for under-represented groups, initiatives that exist & what we can do to be as inclusive as possible.
1
12
60
Putting finishing touches on next week's episode of
#DataFramed
, the
@DataCamp
podcast, a wild ride w/
@StatModeling
through statistics, data science, polling & election forecasting. Andy Gelman has been a hero of mine for years & I can't wait to share this w/ you cos it's 🔥.
3
12
59
The AI Engineer Data Literacy Divide
🎙️In this
@VanishingData
clip,
@HamelHusain
argues that the 'AI Engineer' role leads to stagnation in many organizations, due to the way the skills for this role was popularized.
Key points:
✅ ‘data’ and ‘evals’ aren’t tightly coupled to
5
13
59
If you wanna do data science, learn about the principles of tidy data and how to apply them to real-world datasets 🧹
2
7
59
On next week's ep. of
#DataFramed
, the
@DataCamp
pod, I'll speak w/ Gabriel Straub, Head of Data Science & Architecture at the
#BBC
. I'll put humility aside right now & tell you it's a fantastic episode.
2
7
57
All models are wrong, but some do massive societal harm at scale.
0
8
56
In this clip,
@chipro
breaks down her new blog post on building generative AI systems. We explore:
🏗️ Common GenAI platform components
❌ AI failure types: information vs. behavior-based
🔍 RAG and context enhancement strategies
🛠️ Output improvement techniques
🔄 Fine-tuning
1
8
56
48 hrs ago,
@jeremyphoward
called "openAI is not going to make it" on a panel I hosted. Since then, a lot has happened and J has changed his prognosis.
We're going to speak again soon about why. Register below:
1
9
56
In Ep. 50 of
#DataFramed
, our Season 1 finale of the
@DataCamp
pod, I speak with
@mathbabedotorg
, author of Weapons of Math Destruction, about the current lack of fairness in artificial intelligence & much more.
1
14
53
🎙️New
@VanishingData
podcast ep out with
@rasbt
(
@LightningAI
) about building LLMs from scratch!
We cover the entire LLM lifecycle, what type of skills you need to work with them, what type of resources and hardware, prompt engineering vs fine-tuning vs RAG, and more 🤓
2
10
56