Ask HN: Any SaaS that offers embeddable REPL for teaching code?

1 pointsosazuwa5y ago1 comments

I'm CS faculty at a US university. In the COVID era, I've committed to improving the students' online learning experience as much as I can using SaaS products that I can apply towards instruction.

Right now, the coding in my assignments is in Jupyter notebooks and homework assignments. I don't like it because there is a lot of time and friction (especially in setting up environments) between learning the content and practicing it.

I like how code instruction sites like Datacamp and Plurasite have interactive coding environments in the browser that enable the student to practice a concept as soon as they learn it. I'd love to pay a SaaS company to give me the ability to set up these sort of practice sections and embed them on lecture pages on my teaching platform.

I want almost exactly the tool Datacamp released called Datacamp-Light (https://github.com/datacamp/datacamp-light) But that doesn't look like an actual service and they don't support the packages/libraries I need. Repl.it might be the answer, but does so much and I'm not sure it has my use case. OReilly's Katacoda looks about right, but I think its meant to be enterprise software.

Anybody have any advice?

1 comments

Jugurtha5y ago

You have mentioned pretty much all the tools I wanted to mention (Repl.it, Katacoda, etc).

Hesitant plug: We do have a platform, https://iko.ai, we started for our own use to do ML projects for our clients, but I don't know if it's suitable for your case. It has real-time, no-setup, collaborative notebooks with most of the popular packages pre-installed (30GB+ images worth of libraries). They can also install additional libraries right on the notebook.

People can work together on the same notebook, see each other's cursors, changes, and selections. They can execute that collaborative notebook on their own environment.

We use that to train our models, but we also use that for our weekly calls: we write the agenda in the notebook, then be on a call and go over it point by point editing collaboratively, adding snippets of code for proof of concept or to reproduce a bug in the product. This way we have the meeting minutes, with the executable code, and the prototypes in the same place.

The platform does much more than that (training, tracking, scheduling notebooks, deploying models, + real time dashboards for model monitoring, etc...), but you don't need all that. We've onboarded students our colleague was supervising for their final year projects on ML.

j / k navigate · click thread line to collapse