Manish Maheshwari 

Multimodal AI Product Leadership. I work on multimodal AI models pre-training and post-training | Ex-Datadog, Airbnb, AWS, Microsoft
Views are my own.

[Tim Denning @Tim_Denning]

Obsess. For f*ck sakes. You only get one life. Don’t screw it up by being normal. Go all in. Act like a psycho. Let people call you insane. Please. I beg you. Obsession is the path.

[Manish Maheshwari  @Manish_m09]

@krandiash Awesome work team cartesia! Sub 300ms ttft with useful quality is pretty good.

[Karan Goel @krandiash]

We released Sonic-3.5 and Ink-2, the #1 streaming models for text to speech and speech to text you can use in your voice agents today.

New architectures enable new frontiers for speed and quality.

We're now the only provider to have #1 models for both speaking and listening.

[Manish Maheshwari  @Manish_m09]

@zaptrem @wandb @shawnup Hey Ryan, wanted to check if the wandb web app works for you?

[FlexFusion @FlexFusion_]

IN JAPAN, LAZINESS IS CONSIDERED A DISEASE AND PEOPLE TREAT IT WITH THESE
7 METHODS:

[Katherine Boyle @KTmBoyle]

Never. Give. Up.

[Manish Maheshwari  @Manish_m09]

@karimknaebel @_fracapuano @wandb Hi Karim - Can you please DM me more about the issues
- outlier filtering is unreliable
- negative regex for panel search

[Manish Maheshwari  @Manish_m09]

@karimknaebel @_fracapuano @wandb On forking, were you looking for a UI action from an existing run, an SDK/API workflow, or something else? We do have a fork runs feature in preview https://t.co/mo7jPdxytB
Also, would like to learn more about (2) and (3) in a DM. My DM is open.

[Manish Maheshwari  @Manish_m09]

@karimknaebel @_fracapuano @wandb Forking works with offline runs. You have to call finish() on the offline run first before you can fork it though.

[Manish Maheshwari  @Manish_m09]

@karimknaebel @_fracapuano @wandb This is really helpful, thank you, Karim. The UI sluggishness, especially when unzoomed, sounds like the thing we should dig into first.
Could you DM one representative wandb workspace? It would help us reproduce unzoom issue.

[Manish Maheshwari  @Manish_m09]

@_ueaj That’s really helpful, thank you. Could you DM one representative slow wandb workspace? Also, which action, UX feels laggy, would help us reproduce it.

[Manish Maheshwari  @Manish_m09]

@_ueaj Got it, that helps. For #3, it sounds like the useful bit is auto-written run/change descriptions that become searchable metadata. What’s one concrete query you’d expect to work?

[Manish Maheshwari  @Manish_m09]

@_ueaj For #3, what type of descriptions would be useful for you? Are you imagining auto-summaries of runs/failures, or natural-language Q&A over projects or runs using a bot?

[Manish Maheshwari  @Manish_m09]

@_ueaj @_ueaj, sorry it’s been this frustrating. I am a PM at W&B. Could you share the top 3-5 issues, gaps, or missing features that have been most painful for you? What was slow or buggy for you? Rough bullets are good.

[Manish Maheshwari  @Manish_m09]

@vineettiruvadi @_fracapuano @wandb @vineettiruvadi sorry it’s been so frustrating. I am a PM at W&B. Could you share the top 3-5 issues, gaps, or missing features in Wandb that have been most painful for you? Rough bullets are good.

[Manish Maheshwari  @Manish_m09]

@karimknaebel @_fracapuano @wandb @karimknaebel, sorry it’s been this frustrating. I am a PM at W&B. Could you share the top 3-5 issues, gaps, or missing features that have been most painful for you? Rough bullets are good.

[Manish Maheshwari  @Manish_m09]

@postimortem @_fracapuano @wandb @postimortem, this is unfortunate. Can you please share more about the weird UX you experienced?

[Charlie Bilello @charliebilello]

Price Increases over last 5 years...
New Cars: +20%
Groceries: +26%
Family Health Insurance: +27%
Shelter: +29%
Restaurants: +30%
Home Prices: +37%
Electricity: +39%
Dozen Eggs: +39%
Gas Utilities: +42%
Gasoline: +42%
Transportation: +43%
Auto Insurance: +58%
Fuel Oil: +63%
Ground Beef: +68%
Coffee: +105%

[21Rates @21RatesHQ]

👀