Mogtaba Alim

Giving Every Model a Voice
Prev @Amazon AGI, @awscloud, @UHN

[Ziyad Basheer @ziyadbasheer]

Excited to partner with @NelsonXLee and The General Consulting Company. They are extremely talented and fantastic to work with!

[anthony @asusevski]

You know what, if SWE is solved then I’m gonna assume every bug I encounter is intentional and get mad

[Mogtaba Alim @MogtabaAlim]

Notion and Google Sheets. The final bosses of browser bloatware

[Mogtaba Alim @MogtabaAlim]

@MouaidAlim @Cambridge_Uni Welcome back 🫡

[Mouaid Alim @MouaidAlim]

Happy to share that I’ve wrapped up my MPhil working on Flow matching and Diffusion models at @Cambridge_Uni. Thrilled to be back in Toronto!

[Ryan Junejo @RyanJunejo]

We sat 12 frontier models at a Mafia table for 40 games.

Every role, vote, lie, investigation, and night action was logged and scored.

When it came to detection, we found that @Kimi_Moonshot K3 was the best, figuring out the mafia 72.9% of the time while @claudeai Opus 5 was only able to do so 48.8% of the time.

(🧵 below)

[Cohere @cohere]

How do you go from jailbreaking your PlayStation to CEO of Canada's leading LLM provider? You go to the University of Toronto.

"When you go to @UofT, you get raised into AI." - @aidangomez

[Aidan Gomez @aidangomez]

Canada needs its Canadians. You should come home and help build this country.

[Mogtaba Alim @MogtabaAlim]

Wow! A future with Nvidia as the only AI company??

AI Hardware ✅
Model Training ✅
Model Inference ✅
Model Distribution ✅

They built an omni model. Are they building the omni AI company?

[MTS @MTSlive]

SITUATION DETECTED: Nvidia has agreed to acquire Hugging Face for $12.9 billion, per The Information.

[Amr Kayid 🪼 @amrmkayid]

wooow! Nvidia is playing n-dimensional chess!!

[Mogtaba Alim @MogtabaAlim]

Benchmaxxing is the lazy way to progress.
The best voice AI benchmarks don’t even exist yet.
Embrace the ambiguity, choose a capability nobody measures well, invent the eval, and go build it.

[Hume AI @hume_ai]

We silenced a number in the benchmark audio. It was never spoken.

Some of the highest-scoring speech recognition models wrote it down anyway, in 30–40% of cases.

New research with @huggingface on measuring when a model is optimizing for the test instead of the audio.

[Tarun Sachdeva @tarunsachdeva]

multiplayer ai should be for everyone

not just the people in your slack

[Mogtaba Alim @MogtabaAlim]

@ziyadbasheer Unfortunately true

[Mogtaba Alim @MogtabaAlim]

Speeches in academia are too long.
Commencements, matriculations, keynotes, all of them.
You can’t inspire an audience you lost halfway through a speech

[superwhisper @superwhisper]

Introducing S1-mini ✨

Our first open-weights language model.

A 0.6B parameter model that processes transcripts entirely on your device. Try it in app today.

[Hume AI @hume_ai]

We silenced a number in the benchmark audio. It was never spoken.

Some of the highest-scoring speech recognition models wrote it down anyway, in 30–40% of cases.

New research with @huggingface on measuring when a model is optimizing for the test instead of the audio.

[Amr Kayid 🪼 @amrmkayid]

Such a great story!! 👏

This line really caught my eye:
“From the inside, the market didn’t look finished. It looked like it had barely started.”

That’s exactly how voice AI feels right now. Still so early, so much left to build. What a time to be building in voice 🔥

[Chris J. Maddison @cjmaddison]

So much credibility is lost when folks say 5-10 years to cure all human disease. We can't even measure most of human disease!

[Mogtaba Alim @MogtabaAlim]

@ziyadbasheer @amrmkayid Fire launch 🔥🔥 Congrats!

[Ziyad Basheer @ziyadbasheer]

I'm excited to share what we're been on working on at Aide: making AI agents reliable for production use in customer-facing interactions.

1/ 🧵