apoorva lal
@apoorvalal.com
causal inference, econometrics, ML, arsenal, loud music, unix, FOSS for scientific computing. opinions my own.
apoorvalal.github.io
(passively) maintains @paperposterbot.bsky.social
apoorvalal.github.io
(passively) maintains @paperposterbot.bsky.social
hobbyist programming to build tooling for hobbyist musicianship: glue library for mp3s->stems->midi->chords/tabs using small pre-trained models.
1) trimming is bulletproof since it's just ffmpeg
2) stemming is reasonable
3) midi is hit-or-miss
4) tabs are underwhelming
github.com/apoorvalal/m...
1) trimming is bulletproof since it's just ffmpeg
2) stemming is reasonable
3) midi is hit-or-miss
4) tabs are underwhelming
github.com/apoorvalal/m...
GitHub - apoorvalal/mlodies: glues pretrained models to do stem separation and transcription to help learn music by ear.
glues pretrained models to do stem separation and transcription to help learn music by ear. - apoorvalal/mlodies
github.com
January 10, 2026 at 11:26 PM
hobbyist programming to build tooling for hobbyist musicianship: glue library for mp3s->stems->midi->chords/tabs using small pre-trained models.
1) trimming is bulletproof since it's just ffmpeg
2) stemming is reasonable
3) midi is hit-or-miss
4) tabs are underwhelming
github.com/apoorvalal/m...
1) trimming is bulletproof since it's just ffmpeg
2) stemming is reasonable
3) midi is hit-or-miss
4) tabs are underwhelming
github.com/apoorvalal/m...
if i had a time machine i'd show up at trinity college, cambridge in 1670 and inflict this sentence upon newton
January 9, 2026 at 9:52 PM
if i had a time machine i'd show up at trinity college, cambridge in 1670 and inflict this sentence upon newton
this is an interesting descriptive statistic for bsky network [pareto principle of production, most people are rubberneckers] that is estimable thanks to atproto's openness. script to make your own is attached - make your own!.
what does this look like for other networks? cc @jugander.bsky.social
what does this look like for other networks? cc @jugander.bsky.social
also that topline follower count is an overestimate for the typical people signed up but disappeared / got put off by the general feed and vibe on here reasons. plotted timestamps of last post for my ~5k+ followers and i'd say ~2k are 'active' [in the last 3mo]
code: gist.github.com/apoorvalal/6...
code: gist.github.com/apoorvalal/6...
January 7, 2026 at 4:21 AM
this is an interesting descriptive statistic for bsky network [pareto principle of production, most people are rubberneckers] that is estimable thanks to atproto's openness. script to make your own is attached - make your own!.
what does this look like for other networks? cc @jugander.bsky.social
what does this look like for other networks? cc @jugander.bsky.social
Reposted by apoorva lal
Please feel free to follow this bot along with @paperposterbot.bsky.social for the full range of economics paper on Arxiv.org
January 6, 2026 at 10:44 PM
Please feel free to follow this bot along with @paperposterbot.bsky.social for the full range of economics paper on Arxiv.org
some good games i played in 2025 in no particular order
1. minishoot' adventures
wildly ambitious 2D zelda-like with a spaceship. also wins award for the most unnecessary apostraphy of all time.
1. minishoot' adventures
wildly ambitious 2D zelda-like with a spaceship. also wins award for the most unnecessary apostraphy of all time.
January 6, 2026 at 3:59 AM
some good games i played in 2025 in no particular order
1. minishoot' adventures
wildly ambitious 2D zelda-like with a spaceship. also wins award for the most unnecessary apostraphy of all time.
1. minishoot' adventures
wildly ambitious 2D zelda-like with a spaceship. also wins award for the most unnecessary apostraphy of all time.
'unc' is short for 'unctuous' (excessively flattering, sycophantic). when your chatbot says you're absolutely right, you should yell at it for engaging in unc behavior
January 4, 2026 at 5:07 AM
'unc' is short for 'unctuous' (excessively flattering, sycophantic). when your chatbot says you're absolutely right, you should yell at it for engaging in unc behavior
fun(?) little webapp to plot daylight length D(t) and D'(t) for anywhere on the globe at any time (t) in the year
lalten.org/daylight
sanity checks:
- reproduce solstices as D'(t) crossing y=0
- equator has flat D(t), D'(t)
[cc @dggoldst.bsky.social who frequently posts about D''(t) ]
lalten.org/daylight
sanity checks:
- reproduce solstices as D'(t) crossing y=0
- equator has flat D(t), D'(t)
[cc @dggoldst.bsky.social who frequently posts about D''(t) ]
January 3, 2026 at 5:32 PM
fun(?) little webapp to plot daylight length D(t) and D'(t) for anywhere on the globe at any time (t) in the year
lalten.org/daylight
sanity checks:
- reproduce solstices as D'(t) crossing y=0
- equator has flat D(t), D'(t)
[cc @dggoldst.bsky.social who frequently posts about D''(t) ]
lalten.org/daylight
sanity checks:
- reproduce solstices as D'(t) crossing y=0
- equator has flat D(t), D'(t)
[cc @dggoldst.bsky.social who frequently posts about D''(t) ]
There should be a series of spiritual sequels involving gradually low-octane crimes called "The Altos" (mail fraud,mostly accidental) and finally "The Baritones" (parking over the line).
January 2, 2026 at 8:03 PM
There should be a series of spiritual sequels involving gradually low-octane crimes called "The Altos" (mail fraud,mostly accidental) and finally "The Baritones" (parking over the line).
Reposted by apoorva lal
I uh, made this. It was supposed to be a joke / concept-art thing that scrolls through the torrent of new AI/ML arXiv uploads too fast to read. But I think I iterated too much and made it almost usable.
arXiv AI/ML Catch-Up
Was your New Year's resolution to keep up with arXiv AI/ML preprints? Browse the past week's new uploads in 30 mins.
www.kmjn.org
January 1, 2026 at 11:45 PM
I uh, made this. It was supposed to be a joke / concept-art thing that scrolls through the torrent of new AI/ML arXiv uploads too fast to read. But I think I iterated too much and made it almost usable.
Life cycle of a method
1. scrappy theory
2. killer empirical application (Angrist/Krueger for IV, Card/Krueger for DID, Lee for RD etc)
3. Deep, general theory
4. Review papers
5. Bad empirical applications
6. Empirical reviews
Proximal has skipped 2 entirely. Is this typical for epi methods?
1. scrappy theory
2. killer empirical application (Angrist/Krueger for IV, Card/Krueger for DID, Lee for RD etc)
3. Deep, general theory
4. Review papers
5. Bad empirical applications
6. Empirical reviews
Proximal has skipped 2 entirely. Is this typical for epi methods?
link 📈🤖
Demystifying Proximal Causal Inference (Ringlein, Nguyen, Zandi et al) Proximal causal inference (PCI) has emerged as a promising framework for identifying and estimating causal effects in the presence of unobserved confounders. While many traditional causal inference methods rely on the
Demystifying Proximal Causal Inference (Ringlein, Nguyen, Zandi et al) Proximal causal inference (PCI) has emerged as a promising framework for identifying and estimating causal effects in the presence of unobserved confounders. While many traditional causal inference methods rely on the
January 1, 2026 at 4:58 PM
Life cycle of a method
1. scrappy theory
2. killer empirical application (Angrist/Krueger for IV, Card/Krueger for DID, Lee for RD etc)
3. Deep, general theory
4. Review papers
5. Bad empirical applications
6. Empirical reviews
Proximal has skipped 2 entirely. Is this typical for epi methods?
1. scrappy theory
2. killer empirical application (Angrist/Krueger for IV, Card/Krueger for DID, Lee for RD etc)
3. Deep, general theory
4. Review papers
5. Bad empirical applications
6. Empirical reviews
Proximal has skipped 2 entirely. Is this typical for epi methods?
This is lovely. Also the ascent in the infancy is no joke: i thought i had a pretty productive 2025 but that's nothing compared to The Agent growing a fully functioning spinal column in 3 months
December 31, 2025 at 9:31 PM
This is lovely. Also the ascent in the infancy is no joke: i thought i had a pretty productive 2025 but that's nothing compared to The Agent growing a fully functioning spinal column in 3 months
tweaked a bit, now installable via the chrome webstore
chromewebstore.google.com/detail/math-...
chromewebstore.google.com/detail/math-...
December 31, 2025 at 5:00 AM
tweaked a bit, now installable via the chrome webstore
chromewebstore.google.com/detail/math-...
chromewebstore.google.com/detail/math-...
finally got around to watching the Odyssey trailer; yeah Nolan's anachronism really rankles.
December 29, 2025 at 11:10 PM
finally got around to watching the Odyssey trailer; yeah Nolan's anachronism really rankles.
vibecoded a little chrome extension to fix a perpetual bugbear with reading pdfs: being sent on search expeditions to figure out where a particular parameter was introduced.
Instead, just type in the latex expression and get back unicode that you can ctrl/cmd+F easily
github.com/apoorvalal/m...
Instead, just type in the latex expression and get back unicode that you can ctrl/cmd+F easily
github.com/apoorvalal/m...
December 28, 2025 at 7:09 PM
vibecoded a little chrome extension to fix a perpetual bugbear with reading pdfs: being sent on search expeditions to figure out where a particular parameter was introduced.
Instead, just type in the latex expression and get back unicode that you can ctrl/cmd+F easily
github.com/apoorvalal/m...
Instead, just type in the latex expression and get back unicode that you can ctrl/cmd+F easily
github.com/apoorvalal/m...
Caterpillar of Daughter
December 24, 2025 at 9:24 PM
Caterpillar of Daughter
You can't engineer around truly radioactive founder effects: evidence from the rise and fall of bsky
(Tolerable semi-public social media, for all its faults, was nice while it lasted. Made parasocial and IRL friends, learned a lot. Now everyone's back on whatsapp and discord boards)
(Tolerable semi-public social media, for all its faults, was nice while it lasted. Made parasocial and IRL friends, learned a lot. Now everyone's back on whatsapp and discord boards)
Deleted a post that was going viral and bringing an undesirable amount of attention from anonymous accounts. But, I'm going to be honest, the toxic folks here have won. So much of the usage of the site is about avoiding their attention. I'll just use this site to get book recs going forward
December 22, 2025 at 1:08 AM
You can't engineer around truly radioactive founder effects: evidence from the rise and fall of bsky
(Tolerable semi-public social media, for all its faults, was nice while it lasted. Made parasocial and IRL friends, learned a lot. Now everyone's back on whatsapp and discord boards)
(Tolerable semi-public social media, for all its faults, was nice while it lasted. Made parasocial and IRL friends, learned a lot. Now everyone's back on whatsapp and discord boards)
incredible idea
dosaygo-studio.github.io/hn-front-pag...
dosaygo-studio.github.io/hn-front-pag...
December 19, 2025 at 9:43 PM
incredible idea
dosaygo-studio.github.io/hn-front-pag...
dosaygo-studio.github.io/hn-front-pag...
economics of tech this week is firms repeatedly rediscovering that the optimal price is the inverse hazard rate of the customer valuation distribution plus markup, and that their prior over this is wrong (llms, snap, github pricing runners)
read/tinker: pricingexperiments.streamlit.app
read/tinker: pricingexperiments.streamlit.app
December 17, 2025 at 9:31 PM
economics of tech this week is firms repeatedly rediscovering that the optimal price is the inverse hazard rate of the customer valuation distribution plus markup, and that their prior over this is wrong (llms, snap, github pricing runners)
read/tinker: pricingexperiments.streamlit.app
read/tinker: pricingexperiments.streamlit.app
Reposted by apoorva lal
MIT's famous 'Missing Semester' course (for the Independent Activities Period in January is back.
missing.csail.mit.edu/2026/
missing.csail.mit.edu/2026/
December 15, 2025 at 6:42 PM
MIT's famous 'Missing Semester' course (for the Independent Activities Period in January is back.
missing.csail.mit.edu/2026/
missing.csail.mit.edu/2026/
it would be a genuine novelty if a billboard on the 101 in the bay area tried to sell me a burger / soda / ambulance-chasing legal advice
December 14, 2025 at 5:00 AM
it would be a genuine novelty if a billboard on the 101 in the bay area tried to sell me a burger / soda / ambulance-chasing legal advice
absolutely terrible attempted punny title aside, this is a very funny fig 1
December 12, 2025 at 5:01 PM
absolutely terrible attempted punny title aside, this is a very funny fig 1
based on the amount of navel-gazing "what truly is a science" going on on my feed, i'm assuming it is finals period?
December 10, 2025 at 10:56 PM
based on the amount of navel-gazing "what truly is a science" going on on my feed, i'm assuming it is finals period?
relatedly: tabular foundation models are getting very good.
lalten.org/pages/fmXgbB...
lalten.org/pages/fmXgbB...
December 6, 2025 at 4:34 PM
relatedly: tabular foundation models are getting very good.
lalten.org/pages/fmXgbB...
lalten.org/pages/fmXgbB...
Reposted by apoorva lal
Want to be an intern at Microsoft Research in the Computational Social Science group in NYC (Jake Hofman, David Rothschild, Dan Goldstein)
Follow this link and do your thing! Deadline approaching soonish!
apply.careers.microsoft.com/careers/job/...
Follow this link and do your thing! Deadline approaching soonish!
apply.careers.microsoft.com/careers/job/...
Research Intern - Computational Social Science | Microsoft Careers
Research Interns put inquiry and theory into practice. Alongside fellow doctoral candidates and some of the world's best researchers, Research Interns learn, collaborate, and network for life. Researc...
apply.careers.microsoft.com
December 5, 2025 at 11:35 PM
Want to be an intern at Microsoft Research in the Computational Social Science group in NYC (Jake Hofman, David Rothschild, Dan Goldstein)
Follow this link and do your thing! Deadline approaching soonish!
apply.careers.microsoft.com/careers/job/...
Follow this link and do your thing! Deadline approaching soonish!
apply.careers.microsoft.com/careers/job/...
if you do even moderately computationally demanding tasks, get a GPU - even an older one.
testing scripts and benchmarks on a 5070 series here, 2060 from @detectiveenters.bsky.social in replies, and 1070 here apoorvalal.github.io/lalgorithms/...
testing scripts and benchmarks on a 5070 series here, 2060 from @detectiveenters.bsky.social in replies, and 1070 here apoorvalal.github.io/lalgorithms/...
December 5, 2025 at 2:18 AM
if you do even moderately computationally demanding tasks, get a GPU - even an older one.
testing scripts and benchmarks on a 5070 series here, 2060 from @detectiveenters.bsky.social in replies, and 1070 here apoorvalal.github.io/lalgorithms/...
testing scripts and benchmarks on a 5070 series here, 2060 from @detectiveenters.bsky.social in replies, and 1070 here apoorvalal.github.io/lalgorithms/...