Loading…

Joe Steinbring

Artificial intelligence

Machine-learning libraries, local inference, distributed models, and agentic development tools. It leans on Python from Languages & Frameworks for the libraries and on the vector databases under Databases for retrieval.

I served on an AI working group at Bader Rutter, where I built a retrieval-augmented generation system for the agency.

Explore artificial intelligence

PyTorchI have used PyTorch for neural-network experimentation and machine-learning workflows in Python.scikit-learnI have used scikit-learn for practical machine-learning tasks including preprocessing, modeling, and evaluation.OllamaI introduced Ollama at Bader Rutter as part of getting the agency to explore local AI, and wrote guides for integrating it into Node.js and Python.ExoI started exploring Exo because, unlike Ollama, it can cluster hardware together with RDMA and tensor parallelism, so the memory a model can use effectively becomes expandable.Google AntigravityI have used Google Antigravity as an agentic development platform for planning, implementing, and verifying software tasks.GooseI went looking for an alternative after liking Google Antigravity and working out that it would cost more than I wanted to spend. Since the summer of 2026 I have been experimenting with Goose against local models served by my Exo cluster.

From the blog

Related writing

Articles from JWS News and DEV that relate to artificial intelligence.

  1. How to build a social media bot that can describe an image for you

    Previously, we have looked at how to use Llama 3.2 Vision and Postman to describe an image and how to use Llama 3.2 Vision and Node to describe an image. In today’s post, I want to build a social media bot as a front-end for the same Llama 3.2 Vision model. In August, we looked…

  2. How can you use Llama 3.2 Vision and Node to describe an image?

    In today’s article, we look at how to use Node.js, Ollama, and Llama 3.2 Vision to describe an image. Using this, AI can generate alt-text for you.

  3. How can you use Llama 3.2 Vision and Postman to describe an image?

    I spent much of 2024 discussing how to use local artificial intelligence with your projects. My overarching goal is to show that you don’t need ChatGPT or Copilot to use AI with your projects. Mr. Scoops got the best response, but my favorite example was using LLaVA to describe…

  4. Retrieval-Augmented Generation Questions

    Back in July, we examined how to use Culver’s custard API to train llama3:8b on the flavor of the day at Milwaukee-area Culver’s locations. Earlier this week, we discussed enabling the public to query the model through a social media account. Before we move on to the next topic,…

  5. Lets play with even more Milwaukee custard data

    Last month, we looked at how to get the custard “flavor of the day” data for Culver’s and Kopp’s. We focused on those two shops because there were APIs available for them. Most custard shops won’t have that. A lot of the smaller shops barely even have a website. Back…

  6. Lets play more with Milwaukee custard data

    Earlier this week, we used two non-public APIs to pull in the flavor of the day data for Culver’s and Kopp’s. Back in May, we looked at How to get AI to tell you the flavor of the day at Kopp’s and I figured that the next natural step would be to do the same…

From GoToSocial

Related thoughts

Short posts about artificial intelligence, matched by topic hashtags.

  • #artificialintelligence
  • #ai
  • #llm
  • #rag
Finding related thoughts…