Paritosh Mohan

VP Eng @ | 2x Entrepreneur | Ex - @AWS

@GeorgiaTech alum.

[TwelveLabs (twelvelabs.io) @twelve_labs]

We raised $100M in Series B funding to build what comes next for video intelligence.
Thank you to our investors, customers, partners, and team for helping us reach this milestone.
The road to Video Superintelligence starts here.
#TwelveLabs #VideoAI

[TwelveLabs (twelvelabs.io) @twelve_labs]

Most video models can describe what they see.

Pegasus 1.5 does something more useful for production: it turns video into structured, time-coded data.

Today, Pegasus 1.5 is generally available. 🧵

[TwelveLabs (twelvelabs.io) @twelve_labs]

We rebuilt our video indexing system from scratch—not because it was broken, but because the hidden costs were compounding. Onboarding took weeks, changes required cross-service coordination, and new deployments meant rewrites. The real bottleneck wasn't compute—it was cognitive load. 🧵

[TwelveLabs (twelvelabs.io) @twelve_labs]

At #AWSreInvent, our CEO & co-founder Jae Lee took the keynote stage to share how TwelveLabs is making the world’s growing volume of video truly usable.
With video-native foundation models - Marengo for multimodal search and Pegasus for deep analysis - we’re transforming video from static storage into searchable, actionable knowledge at scale.
Excited for what’s ahead in video AI.
#TwelveLabs #VideoAI

[TwelveLabs (twelvelabs.io) @twelve_labs]

Marengo 3.0 is live, built for real production video workloads, not stitched-together image/audio hacks. Try it on @ProductHunt + share feedback: https://t.co/fzfAK09pjr
What’s new for builders:
⚡ 50% smaller embeddings
🚀 2x faster indexing
⏱️ Temporal + spatial reasoning (not just frames)
🏅 Entity-level search (players, objects, roles)
🖼️➕💬 Image + text queries
🌍 Multilingual + 4-hour video support
#TwelveLabs #VideoAI

[TwelveLabs (twelvelabs.io) @twelve_labs]

Specialization typically degrades general performance. Not here.

☑️MSRVTT (video retrieval): 72.5%
☑️SomethingSomethingv2-MC (classification): 88.2%
☑️UCF101: 93.3%

Marengo 3.0 maintains SOTA on established benchmarks while adding specialized capabilities competitors can't match.

[TwelveLabs (twelvelabs.io) @twelve_labs]

Introducing Marengo 3.0: production-grade video foundation model that delivers 30x faster latency than competitors with 78.5% composite performance across video, audio, image, and text.

This isn't about better benchmarks. It's about the difference between models that work in production and those that don't. 🧵

[Paritosh Mohan @ParitoshMohan]

It's deeply frustrating for a few seconds when I switch from Cursor to any other tool where I have to write. Like, why can't Gmail just complete my thoughts for me?

[Amjad Masad @amasad]

The deflationary impact on software is profound and accelerating.

[Paritosh Mohan @ParitoshMohan]

With each open source model release (this time DeepSeek-V3-0324), the 2023 Google internal memo "We Have No Moat, And Neither Does OpenAI" feels increasingly prescient.

[Paritosh Mohan @ParitoshMohan]

New Beginnings, by @ParitoshMohan https://t.co/AUOY2PmxuU

[Paritosh Mohan @ParitoshMohan]

I just started a Substack. You can subscribe to it here https://t.co/gGZsC1M0hy

[Jan Rosenow @janrosenow]

Low electricity prices are a prerequisite for high levels of electrification.

There are no high electrification regions with high electricity prices.

Great chart from @RockyMtnInst

[Andrew Ng @AndrewYNg]

The effort to protect innovation and open source continues. I believe we’re all better off if anyone can carry out basic AI research and share their innovations. Right now, I’m deeply concerned about California's proposed law SB-1047. It’s a long, complex bill with many parts that require safety assessments, shutdown capability for models, and so on.

There are many things wrong with this bill, but I’d like to focus here on just one: It defines an unreasonable “hazardous capability” designation that may make builders of large AI models liable if someone uses their models to do something that exceeds the bill’s definition of harm (such as causing $500 million in damage). That is practically impossible for any AI builder to ensure. If the bill is passed in its present form, it will stifle AI model builders, especially open source developers.

Some AI applications, for example in healthcare, are risky. But as I wrote previously, regulators should regulate applications rather than technology.
- Technology refers to tools that can be applied in many ways to solve various problems.
- Applications are specific implementations of technologies designed to meet particular customer needs.

For example, an electric motor is a technology. When we put it in a blender, an electric vehicle, dialysis machine, or guided bomb, it becomes an application. Imagine if we passed laws saying, if anyone uses a motor in a harmful way, the motor manufacturer is liable. Motor makers would either shut down or make motors so tiny as to be useless for most applications. If we pass such a law, sure, we might stop people from building guided bombs, but we’d also lose blenders, electric vehicles, and dialysis machines. In contrast, if we look at specific applications, like blenders, we can more rationally assess risks and figure out how to make sure they’re safe, and even ban classes of applications, like certain types of munitions.

Safety is a property of the application, not a property of the technology (or model), as @random_walker and @sayashk have pointed out. Whether a blender is a safe one can’t be determined by examining the electric motor. A similar argument holds for AI.

SB-1047 doesn’t account for this distinction. It ignores the reality that the number of beneficial uses of AI models is, like electric motors, vastly greater than the number of harmful ones. But, just as no one knows how to build a motor that can’t be used to cause harm, no one has figured out how to make sure an AI model can’t be adapted to harmful uses. In the case of open source models, there’s no known defense to fine-tuning to remove RLHF alignment. And jailbreaking work has shown that even closed-source, proprietary models that have been properly aligned can be attacked in ways that make them give harmful responses. Indeed, the sharp-witted @elder_plinius regularly tweets about jailbreaks for closed models. Kudos also to Anthropic’s @cem__anil and collaborators for publishing their work on many-shot jailbreaking, an attack that can get leading large language models to give inappropriate responses and is hard to defend against.

California has been home to a lot of innovation in AI. I’m worried that this anti-competitive, anti-innovation proposal has gotten so much traction in the legislature. Worse, other jurisdictions often follow California, and it would be awful if they were to do so in this instance.

SB-1047 passed in a key vote in the State Senate in May, but it still has additional steps before it becomes law. I hope you will speak out against it if you get a chance to do so.

[Original text (with links): https://t.co/MOQqFF6cID ]

[Frank Nielsen @FrnkNlsn]

Must-read excellent books with 🆓 PDF available to download from the author/book web pages:

👉 https://t.co/zXELEi8E4W

[Yann LeCun @ylecun]

As long as AI systems are trained to reproduce human-generated data (e.g. text) and have no search/planning/reasoning capability, performance will saturate below or around human level.

Furthermore, the amount of trials needed to reach that level will be far larger than the amount of trials needed to train humans.

LLMs are trained with 200,000 years worth of reading material and are still pretty dumb.
Their usefulness resides in their vast accumulated knowledge and language fluency. But they are still pretty dumb.

[François Chollet @fchollet]

Programming is thinking + syntax. Maybe 90% thinking and 10% syntax. While you can automate away syntax, you can only ever outsource thinking to someone/something that can think. Attempts to outsource thinking to a syntax generator do not end well.

[Paritosh Mohan @ParitoshMohan]

2/ But do they see their hold on the market to be so unassailable, that they’d alienate all developers over and over? I can’t help but wonder if this will backfire at some point when sudden technological shifts happen and they need to place nice with others.

[Paritosh Mohan @ParitoshMohan]

1/ Find it fascinating that Apple continues to defend its position so staunchly here. This is of course a massive revenue stream and having built a differentiated product (and a successful duopoly), it makes sense to squeeze the juice out of the platform

https://t.co/YcTQxpRPaX

[Paritosh Mohan @ParitoshMohan]

Nuclear fusion for the grid is coming much sooner than we think

https://t.co/bpTcZ22XK3