Mastodon: kurtthorn@fosstodon.org. Microscopy and DS/ML. Chief Technology Officer @ArrePath. Tweets are my personal opinion. Pronouns: he/him/his.

Joined November 2016
29 Photos and videos
Very good post about the economics of and investor dynamics in drug discovery/biotech/techbio startups, where the hard problems are, and the fundamental challenges in assessing outcomes and quality.
6 Jan 2025
With the start of my first batch as a visiting partner at @ycombinator, this felt like a good time to write my founder story about our journey at @reverielabs. It's a bit of a long one. ankitg.me/blog/2025/01/06/un…
2
10
1,550
Kurt Thorn retweeted
And yes, school board elections matter! Vote for parents’ #1: SUPRYIA RAY @RayforBOE, and rising education leaders like @jaimenina and @ParagGuptaSF
1
3
468
Kurt Thorn retweeted
Thermal expansion -> a typical problem in time-lapse microscopy at high resolution. Here I measure the axial shift of the focal plane of a Nikon 40x0.95 air objective at ~1.6um/K, or about ~1.5x the depth of field per degree C: github.com/amsikking/Nikon_4…
6
22
110
13,133
Kurt Thorn retweeted
I believe James Earl Jones was the last surviving cast member from DR. STRANGELOVE (his first movie). What a career. RIP.
200
4,861
56,961
1,476,090
Kurt Thorn retweeted
20 Aug 2024
Are you new to Cursor and looking for some tutorial videos? Here is a list of the good ones. First place to start for sure is @hive_echo list of Cursor videos: youtube.com/@echohive/search… More below
9
26
242
30,016
Kurt Thorn retweeted
15 Aug 2024
He failed at the two problems, cache invalidation and naming things.
my friend works in fashion. i set her up with one of my tech homies. this is how it went.
Community note
This is engagement farming, this account is know for engagement farming. Example: This post: x.com/roshanpateI/st… Uses an image from this site: moldguy.ca/improving-indo…
4
12
377
21,054
Kurt Thorn retweeted
13 Aug 2024
Come to San Francisco and hang with us!
Tenure-track faculty search of the @Ucsf_Biochem Department is open for applications! We are looking for creative, innovative scientists asking fundamental questions in any area of biology. Join our vibrant, collaborative, and supportive community! aprecruit.ucsf.edu/JPF05209
4
10
1,831
Kurt Thorn retweeted
10 Aug 2024
I gave the “map of every European City” by itchy feet to Claude and asked it to do one for the US. Given its drawing skills, this was pretty solid, even a couple funny bits. (yes, of course similar things exist, this didn’t seem to be pulling from them)
9
20
220
28,595
Kurt Thorn retweeted
Right now the advantages from AI accrue to workers, not firms. As a result, they can be hard to notice (especially if you don't try to seriously use AI yourself, which is illuminating) These two accounts by coders are instructive: nicholas.carlini.com/writing… erikschluntz.com/software/20…
14
84
686
81,725
Kurt Thorn retweeted
ePhective (San Francisco) is a newly founded #phagetherapy company started by @joeBondyDenomy (@UCSF) & Zemer Gitai (@Princeton). They are currently looking for: 2 #Phage Scientists to lead projects in phage discovery, characterization, & engineering
1
22
46
11,510
PCA is surprisingly hard to beat
1
15
991
30 Jul 2024
It’s funny how moving from academia to industry changes your perspective. Looking for my second industry job I spent a lot of time asking questions about business strategy, to find a company that I thought would be successful. That never would have occurred to me in academia.
Replying to @ScopeKurt
💯 when I joined my current company when it was private and much smaller, the first thing I looked for was who they’d hired already - the concept was cool, but they’d also brought in many pharma veterans who knew how to get out the hammer and tongs and get sh*t done
3
449
30 Jul 2024
Agreed. There’s a reason VCs prefer an A team with a B idea over a B team with an A idea. Execution matters more than you think.
Second this. Pharma prioritizes action and decision-making. Scientific excellence is important, but execution is equally important. The work is iterative. Turn the crank, fail fast, learn from the failure, improve design, repeat. You can be scientifically excellent to a fault.
2
2
9
2,012
Kurt Thorn retweeted
Second this. Pharma prioritizes action and decision-making. Scientific excellence is important, but execution is equally important. The work is iterative. Turn the crank, fail fast, learn from the failure, improve design, repeat. You can be scientifically excellent to a fault.
An observation from the industry side, a lot of companies prefer not to hire academic scientists because they are prone to spending too much time in ideation as opposed to building experimentation. I struggled in my job search early on due to this belief about phds in industry
1
11
84
12,438
Kurt Thorn retweeted
Pretty amazing how well SQL and python have withstood the test of time. Some of us will remember an old chestnut, "I don't know what the future of programming will look like, but will be called Fortran." Fortran is down there with COBOL and has almost fallen off the chart.
25 Jul 2024
The StackOverflow 2024 survey has been released. I always find the gap between desired and admired for languages to be the most interesting. I think it is a great indicator for where the software engineering world is heading.
2
2
1,129
Kurt Thorn retweeted
🌈Behold my electromagnetic spectrum!🌈 I made the spectrum by pouring 46 bottles of nail polish & fluorescent acrylic paint.💅🎨🖌️ #FluorescenceFriday #SciArt #SciComm
2
5
26
1,842
Kurt Thorn retweeted
23 Jul 2024
i've always loved the pair-wise ranking loss aka the weston loss @jaseweston,) and unsurprisingly it has recently become a darling for many LLM aficionados who use it to finetune LM's. (1/5)
1
11
129
27,725
Kurt Thorn retweeted
10 Jul 2024
I stumbled across the most amazing paper today "An Abundance of Katherines: The Game Theory of Baby Naming" Authored by Katy, Kate, Katie, Kate, Katherine (@katherine1ee), Katie (@katielulula), Cathy, and Katie (@ktvank) arxiv.org/abs/2404.00732
18
73
519
84,409
Kurt Thorn retweeted
AT LAST! The Sonnet 3.5 test of The Deep Writer is here. This isn't cherry picked. This is SOTA AI. The most complex Agent system I know of, combined with Sonnet 3.5, 100s of API calls, millions of tokens of thought to write this 28 page short story! Results below:👇🧵
7
4
38
4,099