We'll live on in people's memories and hearts. Thank you.
83 karma · joined March 27, 2020
I was thinking of using langchain's markdown parser: https://python.langchain.com/docs/modules/data_connection/do... but I haven't figured out how to put all the chatbot pieces together.
Thank you for the guide.
We also helped with design and infrastructure support for ssa.gov (launched earlier this year) with a contractor team to try to boil down 60,000 pages to ~30 pages that people tend to use.
// opinions of a former USDS, no longer with the team
Can you elaborate on what in particular on the VA site that causes serious harm to the people it's meant to serve? I'm happy to bring it to the team or if you don't feel comfortable with stating it here, please bring it up to the open source website for va.gov. https://github.com/department-of-veterans-affairs/vets-websi...
As for non-sustainable lifestyle, it differs from person to person. It's an appointed position of 2 years, with one chance of renewal for that reason. It's an unique opportunity to see the government from a position of power, so it's expected for USDSers to try to make it count.
It's not always the case USDS will create further technical debt for contractors to develop sites. During cutovers, consider creating playbooks and provide context in those non-stop meetings to the contractors, and best of all, provide a phone number and an email for emergency.
And some of those people from USDS never left the government. ;)
How do people in the field handle configuration updates with code? For example, if I want to set it up as an identity broker to an idp, I would want that configuration backed by code, reviewed by my team. Is anybody using the keycloak terraform provider https://registry.terraform.io/providers/mrparkers/keycloak/l... in production?
Do people diff the realm json configuration as code and use that instead?
draw.io should be available too.
Back in 2018 when I joined, VA laptops were still running on hard disk drives and the laptops were 6.5lbs. Procurement cycles for it were likely once per decade, and VA was just going through their last batch of shitty laptops; and I wrote all my bash scripts on that running on cygwin.
Personally, I didn't even know of this portal's existence until I signed up for an account and and went through the same process the author went through. I noted the same security theater and passed it over to my USDS team to see if there's anything we can do.
At first glance, simply replacing the front end is a solution, but there are bigger problems here that the author did not try. If one tries to change your bank account number, having typed in it wrong, they'll have to mail in a physical form to have it changed. As the author noted, going over $10k per year will also trigger unintended consequences that can span 8-10 weeks before the money is returned.
There are problems here that goes beyond just replacing the "front end".
If you're just starting out, start with none. Quarantine or reject needs to be carefully monitored over time.
I have to assume that VA department you're talking about is Dept of Veterans Affairs - and if that's the case, you're in luck! You don't have to wait till 2150 for this to happen. I know that we no longer process injury claims on paper probably since 2014 or so. Most of the paperwork filed are shipped, scanned, and processed into digitized forms into a system of record known as Veterans Benefit Management System (VBMS) here at the VA. VA Regional Office staffers can then pull the digital records for further processing.
I supported the VBMS team's effort in migrating to a cloud based environment in 2017/2018 and my team help start the Caseflow/Appeals project, a system used to track appeals cases - which relies on VBMS daily. Caseflow is open source as well.
The list is a nice dream, and it is with the will of those in the federal work force who can break down those silos and bring about a common understanding with urgency that'll make it happen for their project.
Are there anything specific about the formats I should be asking for? I'm assuming there are hundred of different file formats (input and output alike). If I knew more about the file formats, that may make the conversation with the team easier.
The following link does include a Linux download, but with little support - https://www.hec.usace.army.mil/software/hec-ras/download.asp....
Also doing some digging on Github, I did find an reverse engineered version of a python version of RAS via win32com control - https://github.com/mikebannis/rascontrol, there's also projects of varying degree across Github - https://github.com/search?q=hec-ras.
I can't speak for all government agencies, but generally there is an internal CA for hosting internal sites. I remember reading a comment from the Federal PKI guide that these sorts of infrastructure goes back to before 2004.
"Prior to 2004, some agencies had already deployed and invested in their own PKI and CAs. Some of these agencies opted out of migrating to the SSP Program and continued to manage their existing infrastructures. These Federal Agencies Legacy operate one or more CAs that are cross-certified with a Federal PKI Trust Infrastructure CA." - https://fpki.idmanagement.gov/ca/
Here's a very short list of public CA certificates from Treasury and it lists out public key certificate for many other agencies as well. - https://pki.treas.gov/crl_certs.htm
Most federal agencies have their own internal PKI systems, and DoD is probably more unique than others because the infrastructure is bigger, older, and has different regulations governing them.
Most civilian agencies such as VA aside from DoD - should utilize public PKI / public CA for their certificates.
Can you go into specifics of the "wonky certificate settings"? I can probably help you out with that or at least bring it to the attention of the team here at VA.
How does planetscale handle the complexity of DIY classic VPN and ensuring a high availability on those VPN links - and ensuring that a certain amount of throughput can be sustained?
Is there a requirement for planetscale to create a full network mesh between all cloud providers, all regions? I'm assuming that it's more selective because it becomes untenable as more cloud regions pop out requiring (n * (n-1))/2 VPN links where n is the number of cloud regions.
Happy to learn anything I can here. Thanks for the blog post.
I'm an engineer with VA as a part of USDS [https://usds.gov]. As dkhenry stated in the comments, I sent a couple messages to the VA Lighthouse team along with the VA Office of Connected Care team to see they can help support this effort.