Arun Swami

Cooking magic @HyBird_Tech
Robotics🤖 @uclrobotics ♥️ SLAM, ML
🌟Whatever Will Happen Will Also be Good🌟 🙏ARUNACHALA🕉❤️🌟

[parm🔬 @parmita]

Good morning. They’re using GPUs for everything the right way, other than saving you from the disease you will have tomorrow.

This must stop. And you must read this:

AI is our attempt to accelerate everything.

Yet curing disease - the thing that matters most - is stagnating.

We have the money. We have the incentive. We even have goodwill.

Yet biotech is drowning in a novelty crisis they bandage over with patent cliffs, and a predictability crisis born from a failure to translate.

The idea that AI can build the best recommendation algorithms on Earth and not systemically destroy every inefficiency in drug discovery is absolutely ABSURD.

BIO's 2011–2020 dataset puts the overall Phase I-to-approval likelihood at 7.9%.

Think about that.

Getting a molecule to Phase I is a Herculean task. The smartest minds in the world, billions of dollars in acquisitions - and humanity's best bet at curing a disease is 7.9% likely. Even after nearly a decade spent getting a molecule to Phase I.

It is simply a problem of what data is being fed.

No AI genius wants to enter a wet lab. No wet lab biologist wants to stop and build better instruments. No analog engineer has any idea that their exact skillset sits between humanity and curing disease.

So we sit. We suffer. And we assume the pharma bros are right that "AI drug discovery is slop."

No. It is simply uncoordinated.

AI is our one and only bet to solve a problem of this grandeur. People understand that well enough today to invest billions into it.

They don't understand the last bridge remaining well enough to solve it.

The last bridge is the analog layer. The instruments. The disease models. The actual physical measurement of what a drug does to a living human cell.

AI's impact in drug discovery is stonewalled by the hard analog problems we haven't solved yet.

Precigenetics is the bet that makes every AI bet start holding true.

Disease must become solvable by silicon. Humans need to stop getting in the way of machines.

It all starts in analog.

[Harveen Singh Chadha @HarveenChadha]

Anyone who is interested in working at a frontier lab must read this tech report from Nvidia

The data engineering section is amazing and look at the amount of different models they used for synthetic data gen

https://t.co/DHaArFTl10

[Arun Swami @ArunSam3]

@ravithejads @MistralAI @aviTwit3 @sophiamyang @NirantK Congratulations 🚀

[Arun Swami @ArunSam3]

@AskYatharth Thanks for this, going through the same and saw this at the right moment! @MoeHafz

[Arun Swami @ArunSam3]

@amy_tabb She grazes around a nearby mountain & stops at us somedays for food. Yes she usually come around by 6am

[Arun Swami @ArunSam3]

@amy_tabb She came today 🤗

[Arun Swami @ArunSam3]

@amy_tabb What 🤷🏾‍♂️

[Arun Swami @ArunSam3]

@amy_tabb Aw cute! Haven't played with horses yet 😑

[Arun Swami @ArunSam3]

@amy_tabb Nice! We have a beautiful cow that calls us everyday at 6am for food and she behaves like a cute dog ☺️

[Arun Swami @ArunSam3]

@sethnabarro @markvanderwilk @AjdDavison @graphcoreai 🎉 congrats for one of the best places for PhD 🎉 all the best 👍

[Arun Swami @ArunSam3]

@akanksha_atrey @AdobeResearch 🎉Congratulations🎉

[Arun Swami @ArunSam3]

@_krishna_murthy @NVIDIAAI @duckietown_coo @DerekRenderling @FidlerSanja @UMontrealDIRO @Mila_Quebec @MontrealRobots Congrats 🎉

[Arun Swami @ArunSam3]

@amy_tabb Interesting so closed loop control rather than open loop control where the world behaves nice & simple😀

[Arun Swami @ArunSam3]

@amy_tabb As @_jdjones_ suggested kinematic free control is an interesting area of research, particularly for miniature surgical robots!

[Arun Swami @ArunSam3]

@amy_tabb Worked with kuka a lot with an assumption of reliable kinematics to get prior pose for 3D fusion. I can understand why for kuka on a mobile robot & grasping! Silently wondered about the same a lot & thinking I am an idiot, so thank you for asking these questions ❤️

[Andrew Davison @AjdDavison]

@SucarEdgar and @wkentaro_ are presenting NodeSLAM this week at #3DV2020. Learned 3D class models for object-based graph maps which enable real-time object SLAM, or to plan highly accurate grasping. Watch the whole video! Paper at https://t.co/K2JnpLYeHD
https://t.co/SE1bT6GMRK

[Arun Swami @ArunSam3]

We ended up having to cut down on potential schools to apply for.
UK universities are much better and transparent with their application process

[Arun Swami @ArunSam3]

These are very relevant and valid questions. Our major obstacle to apply for grad schools in US was entry fees! We have to pay for GRE and TOEFL + pay for (applications & sending test results)! $1000 was roughly my expenditure for a year of undergrad

[Bryan D Martin @BryanDMartin_]

It's grad school application season. This year I'd like to highlight one of the many difficult and most ignored barriers to pursuing an advanced degree: the financial cost of applying.

a 🧵 with my perspective and some suggestions for departments to alleviate this barrier

1/

[Arun Swami @ArunSam3]

Old Euston ❤️

[London Euston @NetworkRailEUS]

Competition Time - fancy winning a poster of Euston from the past?

We’ve 6 used posters available.

All you have to do is retweet to be in with a chance of winning. Rules in the comments ⬇️

@AvantiWestCoast @LNRailway @CalSleeper @nationalrailenq @networkrail

[Rebecca Shipley @rebeccajshipley]

Very much looking forward to this event, led by our @Health_Eng Global Delivery Group. All welcome, come along!