hamilton VS orchest

In this post, we’ll show how your team can turn any utility function(s) into reusable IPython Jupyter magics for a better notebook experience. As an example, we’ll use Hamilton, my open source library, to motivate the creation of a magic that facilitates better development ergonomics for using it. You needn’t know what Hamilton is to understand this post.

We built an app with it -- https://blog.dagworks.io/p/building-a-lightweight-experiment. You can see the code here https://github.com/DAGWorks-Inc/hamilton/blob/main/hamilton/....

Usually we've been prototyping with streamlit, but found that at times to be clunky. FastUI still has rough edges, but we made it work for our lightweight app.

This library is old news? Is there anything new that they've added that's noteworthy to take it for another spin?

[disclaimer I'm a maintainer of Hamilton] Otherwise FYI Prophet gels well with https://github.com/DAGWorks-Inc/hamilton for setting up your features and dataset for fitting & prediction[/disclaimer].

I had been hearing these pains from Langchain users for quite a while. Suffice to say I think:

1. too many layers of OO abstractions are a liability in production contexts. I'm biased, but a more functional approach is a better way to model what's going on. It's easier to test, wrap a function with concerns, and therefore reason about.

2. as fast as the field is moving, the layers of abstractions actually hurt your ability to customize without really diving into the details of the framework, or requiring you to step outside it -- in which case, why use it?

Otherwise I definitely love the small amount of code you need to write to get an LLM application up with Langchain. However you read code more often than you write it, in which case this brevity is a trade-off. Would you prefer to reduce your time debugging a production outage? or building the application? There's no right answer, other than "it depends".

To that end - we've come up with a post showing how one might use Hamilton (https://github.com/dagWorks-Inc/hamilton) to easily create a workflow to ingest data into a vector database that I think has a great production story. https://open.substack.com/pub/dagworks/p/building-a-maintain...

Note: Hamilton can cover your MLOps as well as LLMOps needs; you'll invariably be connecting LLM applications with traditional data/ML pipelines because LLMs don't solve everything -- but that's a post for another day.

This is me. I drive an open source library Hamilton that people doing time-series/ML work love to use. I'm building a paid product around it at DAGWorks, and I'm after feedback on our current version. Can I entice anyone to:

From a nuts and bolts perspective, I've been thinking of building some reactivity on top of https://github.com/dagworks-inc/hamilton (author here) that could get at this. (If you have a use case that could be documented, I'd appreciate it.)

Most people don't track lineage because it's difficult (though if you use something like https://github.com/DAGWorks-Inc/hamilton to write your pipeline - author here - it can come almost for free).

Otherwise, I'm biased here, but check out https://github.com/dagworks-inc/hamilton - it could be your universal layer that expresses how things should flow, that is orchestration system agnostic, which would make it easy to migrate between systems easily.

You can check out our OSS https://github.com/orchest/orchest

The code example is part of our ELT/data pipeline tool called Orchest: https://github.com/orchest/orchest/

First want to say congrats to the Patterns team for creating a gorgeous looking tool. Very minimal and approachable. Massive kudos!

Disclaimer: we're building something very similar and I'm curious about a couple of things.

One of the questions our users have asked us often is how to minimize the dependence on "product specific" components/nodes/steps. For example, if you write CI for GitHub Actions you may use a bunch of GitHub Action references.

Looking at the `graph.yml` in some of the examples you shared you use a similar approach (e.g. patterns/openai-completion@v4). That means that whenever you depend on such components your automation/data pipeline becomes more tied to the specific tool (GitHub Actions/Patterns), effectively locking in users.

How are you helping users feel comfortable with that problem (I don't want to invest in something that's not portable)? It's something we've struggled with ourselves as we're expanding the "out of the box" capabilities you get.

Furthermore, would have loved to see this as an open source project. But I guess the second best thing to open source is some open source contributions and `dcp` and `common-model` look quite interesting!

For those who are curious, I'm one of the authors of https://github.com/orchest/orchest

Haven't tried it. In its favor, Argo is vendor neutral and is really easy to set up in a local k8s environment like docker for desktop or minikube. If you already use k8s for configuration, service discovery, secret management, etc, it's dead simple to set up and use (avoiding configuration having to learn a whole new workflow configuration language in addition to k8s). The big downside is that it doesn't have a visual DAG editor (although that might be a positive for engineers having to fix workflows written by non-programmers), but the relatively bare-metal nature of Argo means that it's fairly easy to use it as an underlying engine for a more opinionated or lower-code framework (orchest is a notable one out now).

If you’d like to try something for you data workflows that’s vendor agnostic (k8s based) and open source you can check out our project: https://github.com/orchest/orchest

You can find it here: https://github.com/orchest/orchest Convenience install script: https://github.com/orchest/orchest#installation

We automatically provide container level parallelism in Orchest: https://github.com/orchest/orchest

For people in this thread interested in what this tool is an alternative to: Airflow, Luigi, Kubeflow, Kedro, Flyte, Metaflow, Sagemaker Pipelines, GCP Vertex Workbench, Azure Data Factory, Azure ML, Dagster, DVC, ClearML, Prefect, Pachyderm, and Orchest.

Disclaimer: author of Orchest https://github.com/orchest/orchest