This message was deleted.

Hi everyone, I'm Justin. I'm interested in hamilton for lightweight data pipelines that I can deploy for training/serving/LLM scenarios. I'm starting a company in this space, and definitely like the looks of hamilton so far! (Just wanted to give a little more intro after I started asking for things!)

This message was deleted.

This message was deleted.

This message was deleted.

Hey everyone! My name is Tobias, I'm a Data Scientist and currently getting into Hamilton. I really like the framework so far but was not able to find documentation on some of the issues I've had, so thought it might be a good idea to join. 🙂

This message was deleted.

This message was deleted.

Hello everyone 👋 , I’m Simon Cheng, DevRel at Vaunt. https://vaunt.dev/ - Empowering Developer Relations! Recognize, celebrate, and grow your open-source community. 💪 Happy to be part of the community!

This message was deleted.

This message was deleted.

This message was deleted.

Hi everyone, I'm a data engineer/analyst building a data consultancy focused on modular data systems (ETL, data modelling, mostly reporting). Using Ibis so I can work with multiple backends. Using Dagster as orchestrator and wondering where Hamilton fits in. Issues I currently have in mind are documentation + lineage, which I want to share with clients somehow. And testing, for which I haven't come across a solid yet lightweight solution. So I intend to read into the Slack threads and ask a few questions here and there!

Hullo all, I work for a HealthTech company called Oxehealth 1 of my roles is to help with ML Ops, which we use for Object Detection & Sleep Analysis. We do inference on-site using a variety of different hardware: Google TPU, Nvidia T1000, A2 & Jetson OrinNX. We do orchestration with Nomad and Ray . I will post in #C03M34FM058 on where I am with getting this working with Hamilton.

Hello, I'm an engineer working on an opensource social media intelligence platform https://github.com/carverauto/threadr/ . I'm currently building the GenAI part of our app using LangGraph and am a big fan of state machines in general.. also wondering if there is anything better. Looking forward to getting to learn more about Hamilton and Burr.

Hello, I am data scientist working on harm detection on social media platforms at Pyrratech https://www.pyrratech.com/. I am using Hamilton for the first time and enjoying the experience so far. I have one question that I will post in the hamilton-help channel.

Hello, I am a data scientist/engineer working in a manufacturing company with nearly 20 years of experience in build data tools/applications in python. I just came accross hamilton and I really like it so far!

Hello everyone, I am Karthik, Machine Learning Engineer with Master’s in Datascience and 5+ years of work experience across 𝐀𝐮𝐭𝐨𝐦𝐨𝐭𝐢𝐯𝐞 𝐈𝐧𝐬𝐮𝐫𝐚𝐧𝐜𝐞, 𝐁𝐚𝐧𝐤𝐢𝐧𝐠 𝐚𝐧𝐝 𝐅𝐢𝐧𝐓𝐞𝐜𝐡 Domains. Along with model training and fintuning I focus on learning and using scalable systems for productionizing ML Models to solve complex problems using AI and ML. Love to be part of the community and contribute in any which way I can. Feel free to connect with me over Linkedin: https://www.linkedin.com/in/karthikkaiplody/ Looking forward to connecting and learning with you all.

Hi, I'm a geologist with a big copper mining company. I visited DAG Hamilton's booth at the PYCON conference in Pittsburgh earlier this year. The folks were really encouraging and helpful in introducing the tool and asking about my use case. I've got a series of Python scripts that do a lot of sequential data processing (mostly csv files and dictionaries - relatively small data). Last weekend I worked to port one to a Hamilton workflow. Good news: I was really pretty happy with the result, especially the graphviz visualization. Bad news: my code was way more awful than I thought it was (global variables, things bound together that shouldn't be bound together, etc. etc.). The tool helped me clean up my code and get a better workflow, but it was work and very humbling. I want to learn more.

Hi all, I’m a former founder, now a technical consultant, focussed on data science, data engineering and optimisation. I’ve been in this space for 15+ years now. I found out about Hamilton via the kedro channels, I only just start playing with it, and it’s been fun so far. I’m always open to connect with technical folk via LinkedIn: https://www.linkedin.com/in/eliasjw/ 🙂

HI all, I am a Head of Data Engg and Data Platform , Technology enthusiast focussed in Data Engg, Devops, MLOps etc. I am going to explore Hamilton as we are looking to move away from Databricks.

👋 Hello everyone, I am Senior Operations Research Analyst at a small consulting company outside of Philadelphia USA. We have some very strict operating restrictions at my work coupled with some equally outdated legacy software. While modernizing our analytical and simulation processes, I tried just about every python pipeline/workflow package under the sun. I always hit the same snags: the package was too heavy and/or not flexible enough for our use cases. Then I found Hamilton and it was breath of fresh air. It’s lightweight, extendable, and super easy to use! My team is already standing it up for use in our production environment and we are churning out real value. Thanks for the great product!

Hi all. I'm a machine learning engineer focusing on the tooling and modeling side of some new projects them in our company. Hamilton is being used in conjunction with metaflow and pytest. It's great to be able to have Hamilton compute stuff for us in the flow and the tests with minimal boilerplate. One use-case that may appeal to some is to set up automatic pytest fixtures based on hamilton DAG nodes. If that sounds interesting to you, feel free to reach out. The idea is to write a bit of code in

conftest.py

so that that once you defined a hamilton node, you get the corresponding pytest fixtures "for free": you just request it in a test and when the test runs, pytest will ask hamilton to build the requested fixtures automatically. Less boilerplate! We are also using a helper class to inspect cached data from the tests (or any other hamilton runs, but 95% of the time it's from the tests). Less boilerplate! Thanks Hamilton team for a great piece of software.

Hi Hamilton Team, First of all, really nice work. I'm in the middle of trying out Hamilton as a possible framework to guide the refactoring of an internal data analytics project here at NVIDIA. I have spent a few days getting your UI, VSCode, and some test code going. Thought I would say hi while working on this. I'll have some questions later on custom data types in UI, how to best structure conditional flow, and the equivalent of looping.

Hi Hamilton team! 👋 I'm in a similar situation to Mattias above, albeit at Shopify. My use case is refactoring a dataflow that's grown from a Makefile. Main pain points atm are saving runs, and DAG visualization (the DAG itself, but more importantly having some viz into intermediate datasets). Hamilton UI and integration with Marimo notebooks are on the top of my list atm.

Hi All. I am new to Hamilton. My use case is not machine learning but rather converting formulas from spreadsheets to Python code. Hamilton is an option because a formula in one cell depends on the output of a formula in another cell and so on. So there is a DAG effect. It seems I will require to use one function per rule but also pass in Series as opposed to dataframes. Will see how it goes and compare it with using a simple pandas dataframe that is updated.

hello all. i've written a few LLM apps that use a bunch of interdependent prompts and noticed that i keep ending up writing caching code. I went looking for something like Haxl to manage the query caching but Hamilton looks a lot closer for what i actually want to do. https://gist.github.com/mwotton/c383b3a5d37da296a87d3b743f3c61b6 i thought the lack of LLM support would be a problem but it does seem like a general purpose workflow tool works out quite nicely. orthogonality and all that... was wondering, is there anyone actively working on making the cache/result store external to the local machine? is the sort of thing that would be nice to share in a team. regardless, i'm now pretty convinced that inverting the structure of your app by representing prompts as an explicit tree is fundamentally the right way to handle this. every app i've written or worked on misses this, so you can't ask simple questions like "if i change this prompt, how many other results could be affected", because it's all bound up with app code and persistence details.

Hi folks - i’ve been looking for something like hamilton forever and am glad to have finally found it. I do data analytics at a small software company and will be using it to create some pipelines to proof-of-concept our some in-development features. Excited to dig deeper!

Hello there. I've encountered a number of less than optimal data processing pipelines in the past and found myself having to debug and optimize them. With some of the steps taking a considerable amount of time, a pattern that was repeatedly in demand was some form of data persistency, that allows working with intermediate results. Usually this resulted in a lot of boilerplate code that was hard to maintain if you value consistency in behaviour. I've been looking through the documentation and examples for Hamilton during the past days and I appreciate the effort that was put in to showcase the difference between Hamilton and many of the other orchestrators out there. What I find particularly compelling about Hamilton, is that it allows for an explorative development process (e.g. using Jupyter notebooks) for feature engineering, which appears to be rather unique among the tools I've seen. Looking at the relative silence in the projects development, I'm a bit worried about its future, so I joined here on Slack to get more of a feeling for the community.

Hi Hamilton community! 👋 I am representing a real estate startup focused on modernizing property investment management. We're working to help property managers transition from Excel spreadsheets to modern data solutions. Our Context: • Small but growing real estate tech startup • Tech stack: Python/Django API, MySQL (AWS RDS), React/TypeScript frontend, all on AWS with Terraform • Currently refactoring a complex, OOP-heavy cashflow calculation system that's become difficult to maintain and optimize • Planning to develop and integrate AI Agents into our platform and considering using hamilton for that as well (seems like thats the main use case) Why Hamilton: We discovered Hamilton while searching for a framework to guide our rewrite, and it seems to tick all our boxes: • Built-in caching capabilities • Enforces clean, modular structure • Strong typing requirements • Native support for distributed calculations Our Planned Implementation: • Breaking down calculations into ~100s of Hamilton nodes • Each node processing NumPy arrays (dimensions up to 1000s × 100s) • Leveraging parallel execution mode extensively • Using dataloader functionality for PostgreSQL outputs Question for the Community: Does our use case sound like a good fit for Hamilton? We're particularly interested in hearing from anyone who's used it for similar financial/numerical computing at scale. Potential Contribution: We're especially excited about the caching features. Since we run on AWS ECS with potential cross-container traffic, we'd love to implement S3-backed caching and would be happy to contribute this as a PR if there's interest. Looking forward to learning from this community and hopefully contributing back!