Gabriel Stechschulte

New post on stochastic model predictive control using sample average approximation

gstechschulte.github.io/posts/2025-1...

Reproducing Uber's Alternating Direction Method of Multipliers (ADMM) based automated budget allocation system in JAX

gstechschulte.github.io/posts/2025-0...

"Intelligent" systems are those that can control and fairly distribute finite resources across multiple-agents.

gstechschulte.github.io/posts/2025-0...

What's worse: TikTok or ChatGPT brain rot?

Since I have replaced Jupyter with marimo, I couldn't be more happy with @zed.dev. Combining these two has been a very pleasant and productive workflow.

I felt the same way since I never had any interest in building them. I now attempt to learn how they work from the perspective of using this knowledge to improve how I utilize them in my day-to-day work.

I like the theme and font style. But yeah, more statistics please 😉 hah

A nice use case. I wonder if they have tests to ensure the re-written queries by GPT return the same data as before?

A quality of a “good” friendship is if you can sit down and not feel the need to say anything for an extended period of time.

And interestingly, this isn’t constrained to tech. Auditors, lawyers, etc. are all an interface to some more complicated system.

A good interface design is a necessary input to make a good/service economically valuable. If a good interface is scarce, how much are people willing to pay for it?

How Uber Uses Ray® to Optimize the Rides Business Computational efficiency is a significant challenge when scaling solutions to a marketplace as large and as complex as Uber. The running and tuning of the Uber rides business relies on substantial numbers of machine learning models and optimization algorithms. Luckily, many pieces of the system can be treated in parallel. Ray® is a general compute engine for Python® that’s designed for ML, AI, and other algorithmic workloads. We describe how Uber has adopted Ray to enable mission-critical systems. Using Uber’s mobility marketplace allocation tuning system as an example, we found performance improvements of up to 40 times that unlocked new capabilities. It also improved developer productivity by increasing iteration speed, reducing incident mitigation time, and lowering code complexity.

This is a really cool blog showing how Uber is using Ray to speed up the optimizing component (Alternating Direction Method of Multipliers) within their budget allocation system.

www.uber.com/en-CH/blog/h...

This is great. Than you!

For “The Algorithms”. Are you able to block or remove YouTube shorts?

Agreed. Then I think back to a blog/paper/book that was “beautifully” written…with optimization continuing soon thereafter haha.

No, I don't need a bloody copilot for this document, thank you.

Writing is, in no small part, a tool for thinking. If you outsource that element to a machine that cannot think, you shouldn't be surprised if, at the end of the process, neither you nor your reader are any the wiser.

An enjoyable read and I also have to admit noticing "a decay" in my programming and problem solving skills.

nmn.gl/blog/ai-illi...

And engineering hours spent defining and handling schemas after-the-fact 😅

The "...using AI" statement usually results in (insert face palm emoji)

Finished Atlas Shrugged after on-and-off reading for 2 years. It is one of those books that alters your perception of society and that you can’t “unsee” after reading.

I wrote this module during the 2023 Google Summer of Code under the supervision of @tomicapretto.bsky.social and @aloctavodia.bsky.social. Feel free to reach out to me or open an issue with feedback or enhancements.

Very awesome! If you are ever in Switzerland, we should do some trail running together 🏃‍♂️

This is one of the reasons I enjoy probabilistic programming. The thought, design, and iterative workflow that goes into building and validating generative models.

Wrote a short article on creating C callbacks with Numba and calling them from Rust using FFI.

gstechschulte.github.io/posts/2024-1...

autodiff in Rust appears to be happening! 🚀

GitHub issue --> github.com/rust-lang/ru...

Rust project goal --> rust-lang.github.io/rust-project...

Hey 👋 I‘m Gabriel and I work in an IoT lab where I focus on data storage and processing, and optimization and control. Looking forward to interacting with the community here.