Tonny Ouma

Applied AI @AWS. Cloud enthusiast. Opinions are my own.⚒️ blue check

[Marc Brooker @MarcJBrooker]

About six months ago, Jean-Baptiste Tristan mentioned an idea that really caught my imagination: a policy language for AI agents based that can reason about history. About order, time, and rate.

Today, we've made that idea real, in Dogwood: https://t.co/ALV9xUgMw6

[Tonny Ouma @techie4real]

@stephen_wolfram Thank you for showing us who Elise really was a wonderful, loving mom, wife and a brilliant mathematician

[Tonny Ouma @techie4real]

Every time I get pulled into RL projects

[Bastian Hagedorn @HagedornBastian]

Nsight Python 1.0 is here 🎉
GPU performance analysis in a few lines of Python: sweep kernel configurations, collect Nsight Compute metrics, compare implementations, derive custom metrics, and generate DataFrames, CSVs, and publication-ready plots.
https://t.co/9V006bqXeV

[Tonny Ouma @techie4real]

@ranman I am so happy for you. Enjoy as much as you can they don’t stay in that stage for long

[Greg Brockman @gdb]

OpenAI is now available for enterprises on Amazon Bedrock:

[Tonny Ouma @techie4real]

@nickevanjoseph Do you even interview the guy?

[Tonny Ouma @techie4real]

The intuition that MLPs are key-value pairs from a mechanistic interpretability perspective is very interesting.
https://t.co/5dExAXiTO9

[Tonny Ouma @techie4real]

Get to your first working agent in minutes: Announcing new features in Amazon Bedrock AgentCore https://t.co/hQUBuZh6ok via @awscloud

[Tonny Ouma @techie4real]

@trq212 @ericrovner @bcherny This is next level customer obsession, and I love it!

[Aryan @justbyte_]

Be honest, is this relatable??

[Anthropic @AnthropicAI]

Finding 1: The longer models reason, the more incoherent they become. This holds across every task and model we tested—whether we measure reasoning tokens, agent actions, or optimizer steps.

[Tonny Ouma @techie4real]

@clare_liguori Thanks for all the work you folks are doing. I woke up this morning and bumped into strands-vllm ...before that, strands-evals was released without any fanfare

[Cognition @cognition]

Devin Review is currently free and works on public or private GitHub PRs. You can use it in three ways:

1. https://t.co/cXMHbmFi2K
2. Swap github for devinreview in the PR url
3. npx devin-review {pr-link} - run this command inside the repo of the PR you want reviewed

Check out the docs for more details: https://t.co/DrRD3EjF57

[Tonny Ouma @techie4real]

@budapp Unlimited credits sounds good

[Tonny Ouma @techie4real]

@rasbt @ManningBooks Just ordered!! Super excited 😊

[Tonny Ouma @techie4real]

I have spent the past year assisting customers in building production-scale LLM applications. @sh_reya and @HamelHusain have done an excellent job of conceptually explaining the path to production, which involves navigating the "Three Gulfs."

[Hamel Husain @HamelHusain]

Published “A field guide to rapidly improving AI products” with OReilly

https://t.co/byEsVEYZqO

[Tonny Ouma @techie4real]

I think our neighbor knows our “secret” sauce

[Mark Chen @markchen90]

Congrats to DeepSeek on producing an o1-level reasoning model! Their research paper demonstrates that they’ve independently found some of the core ideas that we did on our way to o1.

[Hamel Husain @HamelHusain]

New post re: Devin (the AI SWE). We couldn't find many reviews of people using it for real tasks, so we went MKBHD mode and put Devin through its paces.

We documented our findings here. Would love to know if others have had a different experience.

https://t.co/DDqzoAXKkl