
Sr Applied Scientist@Amazon. UW ECE PhD. Clinical LLMs + ML4genetics. From Calcutta. Views are my own. Moving to bluesky: @smukherjee89.bsky.social
Seattle, WA
Joined August 2011
- Tweets13.1K
- Following1.7K
- Followers1.3K
- Likes18.4K
Pinned Tweet
Our EmbedGEM paper is now published in Bioinformatics Advances.
🧬"EmbedGEM: A framework to evaluate the utility of embeddings for genetic discovery" by @SMukherjee89 et al. presents an innovative tool to systematically assess embeddings in #genetic research, focusing on heritability & disease relevance. doi.org/10.1093/bioadv/vbae1… #genomics
Sumit Mukherjee retweeted
Sometimes I think about a counterfactual world in which the corner of the AI ethics community I used to be part of hadn't dogmatically rejected LLMs as unworthy of constructive engagement—a stance that seemed largely driven by contempt for Effective Altruism and its entanglement with AI. In that world, more people holding that community's values would have joined the major AI companies earlier. Maybe that would have impacted model development, maybe not. Without a doubt, though, as of yesterday, there would be a lot more people with meaningful amounts of money to fund the causes they support, rather than all of it being dumped into EA-forward causes like I suspect is about to happen.
Sumit Mukherjee retweeted
I’m hiring a postdoc at Cambridge again!
Work on privacy-preserving human pangenomes at the intersection of statistics, graph algorithms, optimization, differential privacy & computational genomics.
Please spread the word!
jobs.cam.ac.uk/job/57231/
After 1.5 years at Oracle Health, I'll be joining Amazon as a Senior Applied Scientist in the Special Projects org to work on AI for Health. Yesterday was my last day at Oracle. I'm taking the rest of the month off to enjoy some time with my fam. :)
Sumit Mukherjee retweeted
an underappreciated reason NIH funding is so durable: it's less about Congress believing in science, and more about the jobs it supports in lawmakers' home states and districts.
universities and academic medical centers are often among the largest local employers. a vote to cut NIH is a vote to cut jobs back home.
there's a lesson here for biotech. we spend a lot of time explaining why people should be excited about what excites us (cool science, new modalities). we should spend more time making sure our success shows up in things they already care about: jobs, their communities, affordable medicines, national security.
biotech's political future will depend as much on how widely it shares the benefits of innovation as on how much innovation it produces.
Sumit Mukherjee retweeted
Co-signed.
The era of high-frequency research.
The following are common myths:
1. Humans are not needed to do research
2. Cost is so high that only frontier labs can do research.
3. Somehow ai based research is bad for research communities
The most useful way to think of ai is
**accelerating the timescale of research**
It is similar to the transition from low-frequency research to high-frequency research akin to what happened in trading.
What took years now takes days.
Once it’s known that Euler has blowup, in the next week it was shown that Navier Stokes has blowup. The fluid science is nowhere near done, now, we need to figure out how to get to next steps: forcing removal for example.
Compressing years to days and hours should be the goal.
The @__alpoge__ and @DimitrisPapail of the world, who do fundamental work but are agent native, terminally online and collaborative will do very well.
We need to build tools and technologies to help more scientists make this transition soon.
We also need postAGI scientific institutions that operate at this high frequency: systems of collaboration, credit and funding need to adapt to this speed of light.
It’s an extraordinarily exciting time. We should collaborate to solve humanity’s hardest problems at the speed of light.
Sumit Mukherjee retweeted
Looking for postdocs+PhDs to help lead a large-scale study of the effects of long-term AI use, a collaboration between Berkeley+@TransluceAI+others!
Funding+positions available (postdocs/visiting researchers/Transluce affiliations). Please reshare; application in next tweet!
Twitter truly has become a gross disgusting place. A few years ago, before Elon took over twitter, I would have never imagined that instagram would feel like a more wholesome place than twitter!! Yuck …
1
1
111
Sumit Mukherjee retweeted
Many great ideas in here. The part I dislike the most is the explicit bias against attracting talent pools from across the world. This will be a big mistake IMO. I'm also concerned that there is over indexing on the private sector & AI in general. 1/
Sumit Mukherjee retweeted
In 2 years, my lab will most certainly not be publishing our own work in gate keeping journals or conference proceedings. I'm done. Just working on a stable transition out of this horrible game of massive waste of time, effort & funding. Can't wait.
Sumit Mukherjee retweeted
RASC+: Retrieval-Constrained LLM Adjudication for Clinical Value Set Authoring
Sumit Mukherjee
arxiv.org/abs/2606.23992 [𝚌𝚜.𝙲𝙻 𝚌𝚜.𝙰𝙸 𝚌𝚜.𝙻𝙶]
Beautiful essay by @2plus2make5 !!
New piece in @TheAtlantic!
We always hear that AI will cure cancer, and I would immediately benefit if it did. But I argue that racing ahead on generalist AI models creates unclear benefits for cancer that are outweighed by broader societal harms.
(Gift link in next post)
2
491
Sumit Mukherjee retweeted
Experts have three views on the future of work, each credible but sharply opposed.
Who’s right? In a new paper for @CarnegieEndow & @CEIPTechProgram, I lay out the best arguments made by the alarmed, patient, and excited groups.
🧵On the most important points and what policymakers can do today
Just saying AI will create jobs is insufficient IMO. The jobs that AI creates should pay similar or more to the jobs that AI will displace, without resulting in a net job loss.
Nvidia CEO Jensen Huang says don't listen to CEOs with a god complex on AI (lol, Dario) $NVDA
On AI destroying jobs: "these kind of comments are not helpful .. somehow they became CEOs, you adopt a god complex and before you know it, you know everything"
"ground ourselves to talking about the facts" AI will "generate hundreds of thousands of jobs .. trillions of dollars [to the U.S. economy]"
If the argument is that AI will make jobs easier, then we should be willing to pay people well to do easier jobs. As a society we need to start having this conversation seriously.
1
57
These guys are doing cool stuff 👇🏼
an overarching goal of @SophontAI is to build a universal medical foundation model that captures all patient data into a unified embedding.
this model will then allow us to study + predict how this patient state evolves over time, including disease progression.
finally, the model can predict how interventions (drugs) can guide a diseased patient state back to healthy state
To drive this home: we compared two oracles — one minimizing winner radius each round, one minimizing next-round voter disagreement directly. They produce meaningfully different outcomes. Optimizing for centrality and optimizing to reduce polarization are not the same. (5/)
1
22
I'm a novice in this space, so feedback very welcome. Try out the repo if you want to run your own scenarios.
Repo: github.com/mukhes3/electoral…
#SocialChoiceTheory #ElectoralSystems #Simulation
39
I found that the two objectives are in tension. Among the rules studied, rules that pull voters together tend to pull candidates away from their supporter bases, and vice versa. (3/)
1
27



















