Kshitiz Khanal
kshitizkhanal7.bsky.social
Kshitiz Khanal
@kshitizkhanal7.bsky.social
Applied AI/ML research at NC State University. Focused on transportation, energy, and land-use

kshitizkhanal.com

Author of the Public Interest AI/ML newsletter. https://public-interest-ai.beehiiv.com
Wonder Man has heart and soul, nice change of pace and dorection from other Marvel series. The poetry-off was deep
February 3, 2026 at 9:11 PM
Reposted by Kshitiz Khanal
🚨Our new paper led by Dr. @zialyle.bsky.social looked at how climate change poses risks to US drinking water utilities & if their bonds disclose these risks to investors. Utilities serving 67 million people have high risks, but 36% of their bonds don't mention climate: www.nature.com/articles/s43...
Climate change risk index and municipal bond disclosures of United States drinking water utilities - Communications Earth & Environment
67 million customers across the US rely on drinking water utilities that face higher climate risk than accounted for, which exposes major gaps in climate adaptation and resilience planning, suggests a...
www.nature.com
February 3, 2026 at 1:06 AM
Asking Github copilot in agent mode to do a root cause analysis with search has been very helpful to get it unstuck on Rust projects
February 2, 2026 at 2:30 PM
Reposted by Kshitiz Khanal
This is not true; I beg people read the full paper and especially the study design.

The conclusions mirror my (and many other practitioners') conclusions: if you use AI critically and engage both with the question and the answer, it has a net positive impact on both learning and productivity
January 31, 2026 at 9:38 AM
Reposted by Kshitiz Khanal
New post about why Moltbook freaks me out. tl;dr it's that you can't conflate a stateful agent with the LLM. Too much changes when you add state.

this one should be easier to read, more diagrams and easy to navigate text

timkellogg.me/blog/2026/01...
Stateful Agents: It's About The State, Not The LLM
timkellogg.me
January 31, 2026 at 3:08 PM
Reposted by Kshitiz Khanal
Post by @village11.bsky.social about how to build an autonomous agent

imo this snip right here is the distilled form of the difference between Clawedbot vs the agents @cameron.stream would have us build

www.appliedaiformops.com/p/what-build...
January 27, 2026 at 1:38 PM
Reposted by Kshitiz Khanal
The latest entrant in the coding-agent-constructed web browsers is here, this one by @emsh.cat, and it's REALLY impressive - 3 days of development, 20,000 lines of Rust, no Cargo dependencies and it renders HTML+CSS extremely well
simonwillison.net/2026/Jan/27/...
One Human + One Agent = One Browser From Scratch
embedding-shapes was so infuriated by the hype around Cursor's FastRender browser project - thousands of parallel agents producing ~1.6 million lines of Rust - that they were inspired to take …
simonwillison.net
January 27, 2026 at 5:00 PM
Reposted by Kshitiz Khanal
yeah.. even the AI scene over there feels always unnecessarily angry at something or other, real or imagined
Twitter eventually turned into a 24 hour steam of angertainment. It's no longer social media. It's social destruction.
January 26, 2026 at 6:23 PM
Been laughing at Claude for 5 mins
January 27, 2026 at 1:54 PM
Been working with agentic document processing recently. Finding a reasonably sized powerful model is crucial and is something that doesn't tokenize. Doing it the old way - asking people on Discord.
January 26, 2026 at 2:00 PM
Reposted by Kshitiz Khanal
ProPublica’s Rx Inspector allows you to find the factories where generic prescription drugs are made.

You can now see the previously unreleased data that powers our groundbreaking tool.
ProPublica Publishes Unreleased Data on the Origins of Generic Prescription Drugs
The dataset powers our Rx Inspector tool, which lets you find the factories that made your generic drugs and their FDA inspection track records.
www.propublica.org
January 24, 2026 at 10:00 PM
Reposted by Kshitiz Khanal
I was supporting a project from GDM to use ML to boost the value of wind energy before and during this grid breakdown in Texas in 2021.

deepmind.google/blog/machine...

I would hate to think that all the value we created for renewables was just captured by crypto minors.
January 24, 2026 at 11:01 PM
Reposted by Kshitiz Khanal
Looking forward to attending the @ametsoc.org meeting in Houston! Well depending on the winter storm... 😅

I’ll be sharing some of our work on using extreme event attribution information to support resilience planning (ams.confex.com/ams/106ANNUA...).

If you are at #AMS2026, be sure to reach out!
January 23, 2026 at 12:06 PM
Reposted by Kshitiz Khanal
Multi vector is the only way forward to make retrieval better.

Paper: ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios ( arxiv.org/abs/2601.08620 )
Repo: huggingface.co/vidore
January 21, 2026 at 12:53 PM
Reposted by Kshitiz Khanal
NO
January 20, 2026 at 6:26 PM
Reposted by Kshitiz Khanal
IMHO, you would enjoy listening to this pod about a guy who is trying to decentralize the electricity grid the same way telecom & internet infrastructure were decentralized.
Making the electricity grid work like the internet
Jonas Birgersson joins me to explain how "packet-switching" for electrons can lead to energy abundance and grid resilience.
www.volts.wtf
January 18, 2026 at 7:59 PM
Reposted by Kshitiz Khanal
Kind of genius.
January 17, 2026 at 4:16 AM
Reposted by Kshitiz Khanal
OpenAI's GPT OSS is still insanely underrated as a highly adopted open LLM. Downloads are out of control.
January 12, 2026 at 1:40 AM
Reposted by Kshitiz Khanal
Bigger datasets aren’t always better

MIT researchers developed a way to identify the smallest dataset that guarantees optimal solutions to complex problems.

Paper: What Data Enables Optimal Decisions? An Exact Characterization for Linear Optimization
( arxiv.org/abs/2505.21692 )
Blog:
Bigger datasets aren’t always better
A new MIT system identifies the smallest possible dataset that can be used to optimally solve a complex problem with many potential solutions. This technique could help engineers or scientists solve p...
news.mit.edu
January 12, 2026 at 2:22 AM
Reposted by Kshitiz Khanal
Eve Sedgewick said "what I'm proudest of is having a life where work and love are impossible to tell apart" and I'm going to wear this like a locket
January 11, 2026 at 2:36 PM
Reposted by Kshitiz Khanal
New blog post: Don't fall into the anti-AI hype.

antirez.com/news/158
January 11, 2026 at 10:19 AM
Reposted by Kshitiz Khanal
1/ We found that deep sequence models memorize atomic facts "geometrically" -- not as an associative lookup table as often imagined.

This opens up practical questions on reasoning/memory/discovery, and also poses a theoretical "memorization puzzle."
January 8, 2026 at 8:31 PM
Reposted by Kshitiz Khanal
My piece “AI is not a tool” is out in AI & Society. The tool metaphor is all over AI industry rhetoric that markets visual #genAI as the solution to your creative needs doi.org/10.1007/s001... 1/2

Image credit: Beckett LeClair / betterimagesofai.org / CC BY 4.0
January 8, 2026 at 12:02 PM
Reposted by Kshitiz Khanal
Both the Timelapse and Notebook plugins are now available in the official QGIS plugin repository! You can install them easily by searching them in the QGIS Plugin Manager.

Timelapse plugin:
QGIS: plugins.qgis.org/plugins/time...
GitHub: github.com/opengeos/qgi...
December 30, 2025 at 2:04 PM
Reposted by Kshitiz Khanal
New Strix post about it's memory architecture

timkellogg.me/blog/2025/12...
Memory Architecture for a Synthetic Being
timkellogg.me
December 30, 2025 at 3:02 PM