Kevin Madura

💻 |

[Kevin Madura @kmad]

My AIE talk on @DSPyOSS is up - check it out!

Thanks to @aiDotEngineer for a high-signal, well-run event

[AI Engineer @aiDotEngineer]

🆕 DSPy is (really) All You Need

https://t.co/0y4nMyr1cE

@kmad returns to AIE for a special workshop sure to please the @DSPyOSS fans - a comprehensive overview of DSPy! (our second after @lateinteraction's talk with us at AIEWF)

[Kevin Madura @kmad]

Epic line up right here

[sarah guo @saranormous]

1/ 🔥 @NoPriorsPod x @LatentSpacePod chat with @SatyaNadella at @Microsoft Build. He has the sharpest mental models of any public company CEO I've interviewed.

$MSFT is at its heart still a tools company! Big focus on agentic coding, harness & AI evals. Takeaways:👇

[Shubham Saboo @Saboo_Shubham_]

https://t.co/w7v8Vq1JNN

[Kevin Madura @kmad]

@antoine_chaffin I will have to check that out then!

[Kevin Madura @kmad]

*If* this is confirmed this is fascinating - a team converted Google’s quantum algo ZKP into a benchmark and had agents hillclimb against it, eventually exceeding their results!

[Harold Benoit @harold_matmul]

@lateinteraction it was my idea :)

Using GEPA is a very natural workflow for creating LLM programs. The iteration speed is very quick, and it easily allows researchers to bias the optimization with some priors (usually derived from just looking at the data).

Thanks a lot for the great tool!

[Kevin Madura @kmad]

GEPA and @DSPyOSS seen in the Microsoft MAI Pre-training write up!

[Lakshya A Agrawal @LakshyAAAgrawal]

Excited to see the use of GEPA-optimized LLM judges for data filtering in MAI-Thinking-1 model's pre-training pipeline!

[Kevin Madura @kmad]

@JayD0ubleu @_philschmid It’s demonstrably not a gimmick, you can try it yourself!

[will brown @willccbb]

god i'm so excited to have noah on the team. been trying to get him here for almost a year. his record of innovation at the frontier of algorithms + infra for self-improving ai is honestly insane, and i think his recent work is my favorite yet. idk how he's so chill about it.

[Kevin Madura @kmad]

@dbreunig @lennypruss @trq212 @CAISconf Whoa that’s an elegant way of putting it.

Have been thinking about exactly this topic (how many tasks today are under-specified to be useful)

[swyx @swyx]

12.30pm today on the @Microsoft Build stream

@NoPriorsPod
x
@latentspacepod
x
@satyanadella

Join us! :)

[Kevin Madura @kmad]

@litcapital * that may or may not have valid shares

[Kevin Madura @kmad]

@bclavie @aiDotEngineer See you there! The RLM discourse on here has been spicy so be sure to check out my talk

[Kevin Madura @kmad]

@ThePeshwa @lateinteraction @PrimeIntellect Both, actually. I would switch to the other when I ran out of credits. Opus was nice for the big picture, gpt-5.5 for the execution and diagnostics

[Kevin Madura @kmad]

So /goal is awesome

Over the past few weeks I used @PrimeIntellect to train a 149M late interaction model based on GTE-ModernColBERT-v1 using PyLate, focused on clause extraction from legal contracts.

On the MLEB benchmark it does well for its size: it's the best accuracy-per-parameter open model on the task, 3rd of 17 open-source models, ahead of Google's EmbeddingGemma (308M, 0.829) and the same-size legal peer Free Law ModernBERT (0.764), behind only Qwen3-Embedding-4B/8B (which are 27–53× larger).

The agents love the prime cli. I only used the UI for paying my bill.

[Kevin Madura @kmad]

@antoine_chaffin None taken 😆

It was many /goals over time but… yes this was codex/cc doing the experimentation. I was just the human asking dumb questions.

[Kevin Madura @kmad]

And yes this is another @lateinteraction inspired project

[Kevin Madura @kmad]

HF is here: https://t.co/jzXBbZSffo

MLEB benchmark: https://t.co/jVmbGkqe1U

Please, poke holes, this is a learning experiment.

[Gabriel Lespérance @GabLesperance]

RLMs are so resilient.

Multiple times I've run into bugs in our setups. What's interesting is that those bugs only became apparent after careful trace reviews, because the RLM actually found a way forward despite some broken state.

Truly mind-boggling.

[Kevin Madura @kmad]

@bradenjhancock @CAISconf 100% this was such a refreshing experience - the content, people, events - stellar work by the team