Welcome to Owl Posting!
What is this?
This is a blog called Owl Posting, which covers the intersection of science (mostly the life-sciences, but sometimes the physical ones too) and machine learning. It started in April 2024, and an article is released around every 1-2 weeks. The total word-count of this blog is, circa April 2026, is around 250,000~ words.
Who are you?
My name is Abhi. Here is my X and Linkedin, and I live in NYC.
If you’d like to contact me, send me an email at abhishaike@gmail.com or DM me on Twitter.
I am currently helping fund life-sciences datasets at the OpenAI Foundation, underneath the ‘Public Data for Health’ program.
Previously I did a mixture of engineering and writing at Noetik, which is building foundation models of tumor microenvironments. Before that, I did ML engineering at Dyno Therapeutics, doing protein engineering to improve AAV delivery for gene therapy. And before even that, I did ML research at Elevance Health, building multimorbidity risk stratification models using medical claims data.
Do you invest?
Yes! My investing interest is ‘the future of science’. This includes AI-for-science, but there’s a lot of interesting stuff going on outside of that: exotic therapeutic modalities, vastly better chemical synthesis, improving biomanufacturing feedback loops, and so on. Email me at abhishaike@gmail.com to chat more.
How is this blog written?
First, I stumble across something I find confusing. Say, cancer vaccines, or biosecurity. For some topics, I need to talk to researchers in the field, but for others, I can get by with what is already in my brain. After this, I go through dozens of hours of discussions between me and LLMs, because the LLMs are usually omniscient in ways that humans aren’t capable of being. They are also often wrong about what actually matters, so the chats with the humans are helpful for grounding myself.
Throughout all this, I am chaotically assembling together an essay. Sometimes it ends up needing to be rewritten, or dramatically reorganized over time, but I prefer to just start and deal with the consequences as I go at it.
Not to use a Claude-ism, but LLMs are incredibly load-bearing for this, and it’d be impossible for me to cover the diversity of topics I do without them. How much do I let their language influence what ends up in the essay? At least a little! According to Pangram, the detectable AI-writing in the writing I put out varies from 0-15%, though it is typically on the lower end of that. Pangram is usually directionally correct about this, though some articles are much more human-influenced, and others more AI-influenced, than its ruling may imply.
In other words, you should not read me if it is important for your words to be from pasture-raised humans!
Ultimately, my north star for publication is some optimal point of ‘good writing’ and ‘useful teaching’, while also not taking months to publish. Though the LLMs are not great writers, they are incredibly good at explaining, often producing blocks of text that genuinely do nip at the topic in a way I would’ve struggled to match. And I often reuse those texts, shaving off the rougher edges of it, to insert into essays. It is rare that I accept anything larger than a paragraph from them, but some sentences—and sometimes longer ones—are just too good to pass up.
I realize this is walking a fine line. I care about having a clean internet, and will try to not serve you slop!
Does this blog have categories?
Yes! The blog can be separated into the following categories (alongside the most-read articles in that section).
Primers
These are intended to be long, extensively researched deep-dives into specific scientific topics. I stick to the facts as much as possible, but also offer my own opinion pretty frequently. Some examples:
Arguments
These are intended to have a reasonably strong opinion, with mildly more conviction than my actual opinion. Think of it closer to a persuasive essay than a review on the topic, which my ‘primers’ are more-so meant for. Do Your Own Research applies for all my posts, but especially so with these.
Startups
These are posts that each will focus on a single startup that I think is interesting, and deeply examine their scientific foundations, product, and risks. These will typically be biotech startups, but I may also include deep-tech startups in general.
The ML drug discovery startup trying really, really hard to not cheat
Better antibodies by engineering targets, not engineering antibodies
The unreasonable effectiveness of plasmid sequencing as a service
Fiction
Just fiction stories, almost always ones with a bio-themed tint to them, but sometimes not.
Misc
These are just essays that are neither arguments nor primers nor startups nor fiction. Just stuff I was thinking about and wanted to write something about.
Podcast
For many would-be biology articles, I get halfway through them and realize that it’d be hard to do the topic justice, even with Claude/Gemini/ChatGPT assistance. In those cases, I go out and find the most talented person in the field, and have a long-form conversation with them, all filmed in a studio here in NYC. If you’d be interested in coming on, you should apply to be a podcast guest!
Can AI improve the current state of molecular simulation? (Corin & Ari Wagen, Ep #1)
What could Alphafold 4 look like? (Sergey Ovchinnikov, Ep #3)
How do you make a 250x better vaccine at 1/10 the cost? Develop it in India. (Soham Sankaran, Ep #2)
Art
This is more for fun than anything else, but I think biology as a field could do with better aesthetics. So, every now and then, I’ll do some design work and throw it up to be sold. Much of the art here is broadly nice to look at (in my opinion!), but often contain little details only understandable to people who are in the field, which is a balance I quite like. Here is the link.
And the entirety of my posts (including podcasts) can be accessed via the ‘Archive’ section.
