Engineer at Column here! The difference between us and ALL banking as a service companies (Stripe Treasury, Unit, etc) is that we are the actual underlying bank in the transaction. BaaS companies usually wrap one or several banks (like us) to provide their API's.
But what is the practical difference to the API user? This seems like caring that about how the API is implemented, which is usually not a concern for a SaaS consumer.
Yup. You can sign up today for sandbox access, and then contact us once you are ready to go live and move real money.
Truth is, we probably aren't great as a personal checking account, and thats not really the intended use case. But you are welcome to try us out and see what you can build!
for sub in `az account list | jq -r '.[].id'`; do \
for rg in `az group list --subscription $sub | jq -r '.[].name'`; do \
az group delete --name ${rg} --subscription $sub --no-wait --yes; \
done; done;
>I did uncover a ridiculously destructive approach to abusing Azure Service Principals in CI/CD pipelines that deploy infrastructure in Azure (Confused Deputy problem):
> for sub in `az account list | jq -r '.[].id'`; do \ for rg in `az group list --subscription $sub | jq -r '.[].name'`; do \ az group delete --name ${rg} --subscription $sub --no-wait --yes; \ done; done;
The CI provider giving you an over-privileged SP to play with needs to fix that, sure. SPs start with zero role assignments, so it's particularly egregious that they gave it unnecessary permissions.
(Though, for the CI providers I'm familiar with, you the user would be the one creating the SP and providing it to the pipeline. So making it over-privileged would be your mistake.)
But it's not a Confused Deputy problem when you have a service principal with delete access to all resource groups in all subscription and tell it delete those resource groups. Confused Deputy involves a higher-privileged server forgetting to downgrade its privileges on behalf of a low-prvivilege client. The SP is the client in this case - it was created with high privileges in the first place.
One thing I haven’t seen much mention of is getting things to run on the M1’s neural engine instead of the GPU - it seems like the neural engine has ~3x more compute capacity and is specifically optimized for this type of computation.
Has anyone spotted any work allowing a mainstream tensor library (e.g. jax, tf, pytorch) to run on the neural engine?
George hotz got his "for play" tensor library[a] to run on the Apple Neural Engine (ANE). The results were somewhat dissapointing, however, and currently it only does relu.
I’m surprised no one has mentioned what I see as the biggest failings of notebooks: poor handling of connection loss / re-connection. The kernel will continue to run, but a connection hiccup will often make the notebook UI stop updating (and lose any kernel output).
It seems to me that diffpriv is a nascent area of research that has not yet been bastardized by the business community. The complete opposite of a buzzword.
