Palantir Blog

Probably worth also posting this for context: https://blog.palantir.com/korrektur-wie-das-online-magazin-die-republi…

Disclaimer: I used to work at Palantir. However, I never used the Government product, and never as much as saw a demo of the Government product. I'll talk mostly about the Government product, but much of this goes for the Finance product too. Seems like there's a bit of a misunderstanding of what Palantir does. Palantir Government does not do data mining natively. The best explanation of this is from Palantir themselves: http://blog.palantir.com/2007/12/04/what-do-we-do/. There's impressive algorithms under the hood of Palantir Government, but the insight and analysis is provided by government analysts. Palantir provides software, services, and support. The services are mostly data integration, training, and some custom software built on top of the Palantir Government product. Palantir isn't providing analysis for money, they're providing a product that enables analysts to analyze.

A big assumption with this change is that the "Modular Open Systems Approach" (MOSA) [0] [1] will be adequate for integrating new systems developed and acquired under this "fast track". MOSA appears to be about 6 years old as a mandate [2] and is something that big contractors - SAIC, BAI, Palantir [3] - talk about. But, 6 years seems brand new in this sector. I'd be curious to see if LLM's have leverage for MOSA software system integrations. [0] https://breakingdefense.com/tag/modular-open-systems-architecture/ [1] https://www.dsp.dla.mil/Programs/MOSA/ [2] https://www.govinfo.gov/app/details/USCODE-2016-title10/USCODE-2016-ti… [3] https://blog.palantir.com/implementing-mosa-with-software-defined-syst…

fuck palantir. evil company. and fuck this bullshit: https://blog.palantir.com/pcl/home

Palantir on User-Centered ML: https://blog.palantir.com/user-centered-machine-learning-0514adb4c98b

this data sovereignty issue is almost certainly going to come up with the UK NHS / Palantir Federated Data Platform[0] which is going to be using Microsoft and AWS cloud services. [0] https://blog.palantir.com/palantir-and-the-nhs-dd1362982fa9

No, it makes no difference. Git doesn't look at the full history when performing a merge, it just looks for similarity between the paths and files in the commits being merged, and the common base commit (which you can discover by running `git merge-base`). It doesn't keep any metadata about whether something was a move: as far as the actual data structures are concerned, you deleted file a/b/c and added file d/e/f, that have the same file content hash. You can find more information on how Git now does it at https://blog.palantir.com/optimizing-gits-merge-machinery-1-127ceb0ef2… . I think this first started to be released in v2.31.0, and completed in v2.33.0. v2.34.0 switched the default merge strategy to the new 'ort' strategy mentioned in that blog.

(I work for Palantir but any opinions are my own) Palantir is not in the business of collecting data, and as such would have no use for this data. Here's an official blog post explaining this: https://blog.palantir.com/palantir-is-not-a-data-company-palantir-expl…

On the other hand, they openly and frequently share information about their work on Quora: https://www.quora.com/Palantir-Technologies. Here they sort of address the apparent secrecy: > This is a common misconception. We do have clients whose names we can't share, but we try to be as open as we can, both internally and externally. We put as much of our customer work as we can on our Analysis Blog (http://www.palantirtech.com/gove...). We participate in the open source community (https://github.com/palantir/). We have engaged with press who have pretty thoroughly profiled the company (http://www.washingtonian.com/art...). We share how our technology works on our white videos (Google "white videos"). We even tell people exactly what we look for when we interview candidates and give tips (see entries in (http://blog.palantir.com/) http://qr.ae/pgHRy

A good complement to our other post last week, How to Rock an Algorithms Interview - http://blog.palantir.com/2011/09/26/how-to-rock-an-algorithms-intervie…

> It should be clear that working at a fundamentally harmful company, like Exxon, Marlboro, or Peabody is going to self-select for certain types of people and those people are simply not care about the wellbeing of others as much. I don't really buy this. Have you seen Palantir's commitment to DEI?[0] How about Lockheed-Martin's commitment to the LGBT community?[1] On the flip side, I've heard plenty of people say Intel is a nightmare to work in, and they're (for the most part) fairly neutral in terms of "harmfulness." And hey, Blizzard-Activision just makes video games, certainly they have a positive and welcoming culture that cares about the wellbeing of their employees, right? My point is, company culture is completely orthogonal to the industry a company serves. Pretending that big oil or tobacco employs exclusively people wearing 19th century robber baron outfits holding big sacks with dollar signs painted on them is naive. Everyone below the executive level is a person trying to feed their family before anything else. [0] https://blog.palantir.com/tagged/diversity-and-inclusion [1] https://www.lockheedmartin.com/en-us/news/features/2021/16-years-of-pr…

The markup in this case would be some type of semantic web format, like JSON-LD[0] or OWL[1], or some database that can process SPARQL[2] queries. Goal being the "inverse" of something like OWL2Vec[2]. EDIT1: A few weeks ago, a team of Brazilian researchers published a report about using ChatGPT to enhance agriculture-focused OWL dataset[4]. EDIT2: In addition to training LLMs on ontologies, it looks like Palantir is using ontologies as guardrails to prevent LLM hallucinations[5]. Makes sense. [0] https://json-ld.org/ [1] https://www.w3.org/TR/owl2-syntax/ [2] https://www.w3.org/TR/sparql11-query/ [3] https://arxiv.org/abs/2009.14654 [4] https://arxiv.org/abs/2504.18651 [5] https://blog.palantir.com/reducing-hallucinations-with-the-ontology-in…