“Manage in good times so you’ll do well in the bad times.”
Joined December 2017
- Tweets 1,096
- Following 240
- Followers 12
- Likes 305
Photos and videos
- retweeted
15 Mar 2025
Yuki Tsunoda has now outqualified Liam Lawson for the last 7 consecutive race weekends 👀
Did Red Bull pick the wrong driver?
#AusGP
657
839
17,331
1,005,606
- retweeted
27 May 2024
THREAD
The best moments from this year’s Giro d’Italia.
1. Tadej Pogačar giving the bottle to one young fan. It is about small things which can make somebody a memory for the whole life. 🥰 (🎥: eurosportitalia)
71
659
10,559
4,922,968
- retweeted
20 May 2023
The majority of humanity lives within a relatively small bandwidth of latitude. This #map is one of my all time favourites. It was created by radicalcartography.net in 2008. A true classic! buff.ly/436j6Zd
34
1,021
3,874
431,600
- retweeted
18 Oct 2022
This!
PS: here's some trivia about XGBoost & LightGBM since their diff is a FAQ in class:
XGBoost's trees are based on breadth-first search, comparing diff features at each node.
LightGBM performs depth-first search, focusing on a single feature & growing the tree from there.
17 Oct 2022
The tooling trifecta for ML:
1. Tabular: XGBoost/LightGBM
2. Text: @huggingface Transformers
3. Vision: @wightmanr's timm
I can solve 99% of the problems I come across using those libraries.
7
36
275
👋🍕 @normconf main talk recordings are live here! We’ll link them on site too soon, just wanted to get these out. Enjoy!
youtube.com/playlist?list=PL…
1
32
112
19,429
- retweeted
13 Dec 2022
ICYMI I gave a (very rough) talk on causal forecasting last Friday. Some broader ideas about how DAGs can be ordered in time, and some discussion of this system we built at Lyft: eng.lyft.com/causal-forecast…
11 Dec 2022
Rest of the talk series, including @seanjtaylor on causal forecasting, here
metaculus.com/questions/1385…
1
5
30
- retweeted
13 Dec 2022
10 months in ...
Can't wait for Twitter to increase the char limit so that I can add ChatGPT to that list!
7 Oct 2022
Ok, so 9 months in we got
- DALL•E 2 [6 Apr 2022]
- Imagen [23 May 2022]
- Stable Diffusion [22 Aug 2022]
- Make-A-Video [29 Sep 2022]
- Imagen-video [6 Oct 2022]
Anything missing?
3
5
48
- retweeted
28 Nov 2022
Next to confirming our previous aggregate results, this in-depth evaluation sheds more light on the directions to improve TabPFNs, and we’re excited to push this further. Stay tuned🙂
All code & plots for this evaluation are available here: colab.research.google.com/dr… 17/17
1
10
- retweeted
25 Nov 2022
Nixtla is building time series forecasting tech the right way: focusing on evaluation/reproducibility, playing well with engineering systems, and providing a variety of options so users can choose the right tool for the job.
25 Nov 2022
@nixtlaInc started last year as a side project
Today we reached 1 million downloads🎉
Our goal is to shake the time series industry and make state-of-the-art algorithms available for everyone 🔥
This is how we got there
🧵1/11
#reinventforecasting
5
24
- retweeted
24 Nov 2022
explainpaper.com allows you to upload any academic paper, highlight a text which results confusing to you and get a clear explanation. Bonus: you can ask follow up questions if you need a more in-depth answer
33
324
1,929
- retweeted
20 Nov 2022
For practically every role, it contains a comprehensive database.
kickresume.com/en/help-cente…
105
470
2,468
- retweeted
12 Nov 2022
I ❤️these slides from @mgershoff about Bayesian vs. Frequentist analysis that concludes:
"This is second order issue.... What you are testing, design of experiment, and execution are all an order of magnitude more important than form of analysis."
drive.google.com/file/d/13Ab…
1
6
46
- retweeted
12 Nov 2022
I started my career in Data Science back in 2016.
Here's a list of Tech YouTube Channels I've ardently followed!! ✅
A Thread 🧵👇
59
774
2,564
- retweeted
29 Oct 2022
Deep Learning Course, University of Geneva
A great deep learning intro course that covers various neural network architectures(ConvNets, MLPs, RNNs, attention, generative models) & techniques for training them. Really good notes & slides!
DL course: fleuret.org/dlc
21
287
1,191
- retweeted
21 Oct 2022
This may revolutionize data science: we introduce TabPFN, a new tabular data classification method that takes 1 second & yields SOTA performance (better than hyperparameter-optimized gradient boosting in 1h). Current limits: up to 1k data points, 100 features, 10 classes. 🧵1/6
104
731
4,089
- retweeted
23 Sep 2022
This might well be the cleanest and clearest PyTorch coding tutorial I've seen so far. Double BAM!!
19 Sep 2022
Coding Neural Networks doesn't have to be frightening...all you need is PyTorch Lightning!!! BAM!
youtu.be/khMzi6xPbuM #BAM #STATQUEST
3
57
444
- retweeted
20 Sep 2022
Our understanding of MLOps is limited to a fragmented landscape of thought pieces, startup landing pages, & press releases. So we did interview study of ML engineers to understand common practices & challenges across organizations & applications: arxiv.org/abs/2209.09125
28
325
1,678
- retweeted
18 Sep 2022
ETNA - Time Series Transforms, Feature Engineering and Predictions
Python GitHub:
github.com/tinkoff-ai/etna
Documentation:
etna-docs.netlify.app/
2
36
213