peterschmidt

Hi everyone, I'm the creator of dstack, a tool that makes it easier to train models in the cloud. Our tool allows extending it with custom providers to support different languages, frameworks, etc.

All the built-in providers are also open-source: https://github.com/dstackai/dstack

Today, we've released a new update that extends the capabilities of dstack beyond training models, and now also allows users to quickly build and share apps with Streamlit, Gradio, and FastAPI in the cloud – in just a few clicks.

Also, similar to apps, it's possible to run dev environments with the required hardware and data access also in one command from the Terminal. All you have to do is to link your own AWS account to run commands.

Currently, we support VS Code, and JupyterLab but plan to support more.

Invite everyone to give it a try and share their thoughts. Happy to discuss the approach and what would be great to have in terms of other features!

Blog post: https://blog.dstack.ai/introducing-apps-and-dev-environments

P.S.: Currently, it's possible to run models and apps only in the configured cloud. If you'd like the tool to also allow you to run it locally, and if you would like this part to be open-source too, please leave comments!

We are building an open-source software that can run on-premises and must provide a way to authorize users. Now we need to make a decision of what first tools for authorization we want to support first. Please share your thoughts on what tools should be supported among the first.

Our target users are medium and large -sized enterprises.

Should it be Google, SAML, LDAP, Active Directory, all, or something else?

Appreciate your help.

I'd like to understand what tools and practises data analysts and data scientists (if any) normally use to track and share reports (either ad-hoc reports or regular reports) with the team (or managers).

Any advice?

I work at a product team in a relatively large SW company. To continuously improve the product experience, we wanted to get more complex insights out of the data. Because we were not getting it from enterprise solutions such as Google Analytics, we decided to build a data pipeline and tools for data analytics for the company.

While building it for ourselves with a great team (data engineers, DevOps), I learnt that such internal product is only affordable for large tech companies who have the resources to ramp up their own team of experienced people. With some market research, I came to the conclusion that there is a need for a lightweight and relatively cheap SaaS for small and medium sized teams to offer data services.

Right now, I have teamed up with a friend to see if we can build a SaaS business around the problem. We are currently engaged in building an MVP and validating the idea. The MVP shall offer some of these

- An API for reporting product logs

- An infrastructure to collect and process product logs

- A service to access the raw data for exploratory analysis

- A service to govern the data (validation, privacy, etc)

- A service to manage A/B experiments

- A service to manage ML models.

- A marketplace for third-party add-on services to work with data

In short, we want to offer services that helps small teams with one or two people capable of working with the data (through pandas, NumPy or Spark) can easily gain deep insights about their products without going through hassles around data infrastructure and data tools.

The current market segment we want to focus is mobile app (Android, iOS) makers.

At this point of time, we are looking for evidence that proves or disproves our idea. We are also looking for early adopters who are willing to use our MVP and give us feedback.

I would deeply appreciate if the HN community can give us any feedback / insights on the above stated topic.