Prabhal Machahary (@prabhalmac) header
Prabhal Machahary (@prabhalmac) profile picture

applied ai engineer, building @MirikaLabs

Joined March 2026
Photo by Prabhal Machahary (@prabhalmac)Photo by Prabhal Machahary (@prabhalmac)Photo by Prabhal Machahary (@prabhalmac)Photo by Prabhal Machahary (@prabhalmac)Photo by Prabhal Machahary (@prabhalmac)Photo by Prabhal Machahary (@prabhalmac)Photo by Prabhal Machahary (@prabhalmac)Photo by Prabhal Machahary (@prabhalmac)Photo by Prabhal Machahary (@prabhalmac)Photo by Prabhal Machahary (@prabhalmac)
Prabhal Machahary retweeted
The idea of an intelligence explosion caused by recursive self improvement has been around for a long time but until very recently it did not seem imminent. Now many leading researchers think it may happen quite soon. You can read our paper about it here: casp.ac/reports/intelligence…
337
1,429
218
6,621
822,804
Prabhal Machahary retweeted
Oct 8: Life Biosciences will present early ER-100 data in glaucoma.
17
31
4
266
28,116
Prabhal Machahary retweeted
Many drugs work by binding to a specific target in the body and blocking or changing what it does. An important first step in the drug development process is designing a molecule that can bind tightly to its target. Traditionally, that's meant weeks or months of expert work per target, sifting through a large number of candidates to identify the few that work. We wanted to test if Claude could successfully design novel protein binders from scratch (also called de novo design). With a protein design prompt written by a human expert, Claude autonomously designed protein binders against 14 out of 15 targets. We then worked with Adaptyv Bio and Twist Bioscience, who independently built and tested the proteins Claude designed.
599
1,432
671
12,419
4,355,424
Prabhal Machahary retweeted
Releasing the model weights and technical report of Kimi K3. Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window. New model architecture: 2.5x the intelligence per unit of compute, not just more params. Alongside Kimi K3, we're opening up more of the stack behind it — high-performance attention kernels, MoE communication library, and infrastructure for running agent environments at scale. Model weights: huggingface.co/moonshotai/Ki… Tech report: github.com/MoonshotAI/Kimi-K… Tech blog: kimi.com/blog/kimi-k3
1,537
7,159
2,352
45,863
14,758,537
Prabhal Machahary retweeted
BREAKING: Chinese AI startup Moonshot officially releases the open weights for its 2.8 trillion parameter Kimi K3 model.
124
325
40
5,933
368,768
Prabhal Machahary retweeted
21 students have lost their lives. Thousands of people are protesting. And this clown is shamelessly laughing. He must be humbled and brought to justice before he dies.
383
2,435
88
11,593
165,901
Prabhal Machahary retweeted
Demis Hassabis says we're entering the "foothills of the singularity." 👀 - He believes AGI could arrive around 2030 (plus or minus a year), marking the beginning of a completely new era for humanity. - According to Hassabis, this year's advances in AI agents, reasoning, and tool use are the first real signs that AI is becoming genuinely useful in everyday workflows. - He says multiple breakthroughs are now converging at once, making progress much faster than expected and giving him confidence that this transformation is approaching. - Hassabis believes society has only a few years to prepare for the profound impact AGI could have, because the choices we make now will shape the future. - Beyond AGI, he says AI could revolutionize biology and medicine. After AlphaFold predicted the structures of over 200 million proteins, his company Isomorphic Labs aims to reduce drug discovery from years to months and eventually even weeks. The next decade could become one of the most transformative periods in human history.
10
39
5
172
19,142
Prabhal Machahary retweeted
NEO's finally learning sign language
116
257
84
3,177
230,435
Prabhal Machahary retweeted
Announcing Grok 4.5, our first model trained specifically for coding and agents. It was trained with Cursor and offers frontier intelligence at leading speeds and cost efficiency. x.ai/news/grok-4-5
1,485
3,447
1,168
26,448
69,314,344
Prabhal Machahary retweeted
Even if your phone is on 1%, take your time to watch this 13 minutes Ronaldo edit as he retires from international football today. The G.O.A.T. 🐐❤️
121
5,246
90
25,839
292,659
Prabhal Machahary retweeted
Introducing Claude Science, a new app designed with every stage of research in mind. Artifacts traced to their code, environments managed on demand, and 60+ optional scientific databases that you can connect. Available now in beta.
1,069
3,414
1,392
35,327
9,837,864
Prabhal Machahary retweeted
Yes!! Really happy with this one! Enjoyed the battles on track and great to be in the fight again. Strong result from the team @redbullracing 👊 Proud and thankful for all the orange support around the track this weekend 🙏
594
6,951
257
54,362
769,403
Prabhal Machahary retweeted
This is an important breakthrough for computational biology. Researchers at the Shanghai Institute of Organic Chemistry developed Void-X, a generative AI model that predicts protein interactions at the atomic level. Unlike conventional protein design approaches that work from the top down, Void-X takes a bottom-up approach. The model learns how atoms pack together inside proteins and generates missing atomic structures by filling microscopic "voids" in protein interfaces. It's trained on more than 8 million atomic clusters, Void-X achieved 78.3% accuracy on intra-chain structures and 68.2% accuracy on inter-chain protein interactions. The significance is bigger than the numbers guys 👀 Many drugs, antibodies, biological enzymes, and other protein-based therapies depend on precise atomic interactions between proteins. By learning these interactions directly, AI could help scientists design new therapeutics with greater speed and precision. welcome to the fastest accelerating SciTech Era 🫡♥️!
6
25
5
120
10,005
Prabhal Machahary retweeted
The official FIFA World Cup ball went to space! We're working to inspire the next generation by showing how space exploration inspires innovation in sports science — and everyday life. Learn more: go.nasa.gov/43G4z94
977
4,103
314
23,437
2,185,421
Prabhal Machahary retweeted
A technical dive inside our new "Midjourney Scanner"
1,183
3,152
2,388
27,877
11,551,076
Prabhal Machahary retweeted
Announcing a new division of Midjourney called "Midjourney Medical"
2,467
4,971
4,091
39,385
18,972,557
Prabhal Machahary retweeted
Indian telecom Reliance is sabotaging access to Telegram for millions of users OUTSIDE India (including the UAE) via a rogue method called BGP hijacking. The sabotage seems intentional, as Reliance has ignored multiple reports. This may be part of a competitive war, as Reliance is partially owned by Meta — the company behind WhatsApp. Network operators are advised to reject unauthorized BGP announcements from Reliance (AS18101) to prevent route hijacks and ensure stable Internet access for their users. Such abuse of global Internet routing is alarming. I wouldn’t be surprised if Reliance/WhatsApp were also behind the recent lobbying effort to ban Telegram in India.
2,130
8,781
916
50,053
3,366,859
Prabhal Machahary retweeted
India’s IT ministry banned Telegram for one week because some users shared leaked exam questions. This punishes 150M+ ordinary Telegram users in India — not the insiders who leaked the exam materials. And the ban hasn't stopped anything. The leaks just moved to other apps.
Statement : Shutting down Telegram is a band aid solution and is a disproportionate answer to exam fraud The Internet Freedom Foundation objects to the directions announced today in the National Testing Agency's press release on action against the Telegram platform. On the NTA's recommendation, the Ministry of Electronics and Information Technology has, under Section 69A of the Information Technology Act, 2000, restricted access to the whole of Telegram in India until 22 June 2026, and has separately ordered the platform to switch off message-editing for every Indian user until 30 June 2026. This is a blunt, nationwide measure aimed at the conduct of rampant fraud rackets, and on the Government's own admission is constitutionally incompatible. At the outset it is important to note that Section 69A and the Blocking Rules of 2009 framed under it allow the Government to block access to specific “information” on a computer resource. They do not extend to switching off an entire intermediary, still less to ordering a company to redesign its product by removing a feature for a whole country. In Shreya Singhal v Union of India, the Supreme Court upheld Section 69A because it is narrow and hedged with procedural safeguards. Reading it to authorise shutting down a platform that lakhs use is an overbroad restriction by the NTAs own admission. For the message-editing direction the release identifies no source of power at all. If one exists, the order must say so. The release argues against itself A restriction on access has to be the least intrusive measure that achieves its aim as per the constitutional test of proportionality laid down in Justice K.S. Puttaswamy v. Union of India (2017) and applied in Anuradha Bhasin v. Union of India (2020). The NTA's own narration shows the block fails its nodal agency, the release says, “has secured the prompt take-down of a substantial number of Telegram channels, groups and bots”, and this targeted work “is the reason the harm caused by these rackets has been contained to the extent it has”. If channel level takedown contained the harm, the case for a blanket block collapses and hence the Government has reached for a heavier tool while conceding that a lighter one was working. The collateral cost sits on the record too as noted in the press release. The block, the NTA accepts, “affects lakhs of citizens who use the Telegram platform for legitimate personal, educational, professional and informational purposes”. The release also says there is "no such paper available outside the secured examination chain" and that “the security of the examination is unaffected by the action taken”. If the exam is secure and no leak exists, what is being suppressed is rumour, and rumour cannot justify closing a platform when specific blocking and criminal prosecution remain available. Students use of Telegram The block of telegram is reactive and ineffective and will punish ordinary users instead of addressing the systemic source of exam leaks. This blocking comes in the final days of NEET preparation, when thousands of students depend on Telegram for study groups, doubt-clearing, and shared resources. Also, it is important to consider that the source of exam papers leak will occur from inside the system, among insiders and across the printing and logistics chain, with the platform being the most downstream channel for distribution. Hence, switching off Telegram, is merely a deflection from the repeated failures that will continue while media attention is directed towards this Telegram ban. Lack of transparency At present only a press release from the NTA has been provided, which recommended the block but the reasoned order of MeitY, the authority that issued it, has not been released. The Anuradha Bhasin decision requires that orders restricting access be published so they can be tested in court. Here the order, and the reasoning of the committee behind it, stay out of view, and we do not know whether Telegram was heard at all. An announcement of a block is no substitute for an order the affected party can challenge. Blunt to enforce and very easy to evade Usually, app-level blocks run through IS-level DNS and IP filtering. They are over inclusive, sweeping in lawful use, yet simple to evade as a determined exam leak racket moves to a VPN or a mirror within minutes while ordinary users lose the service for a week. We ask the Government to: 1) Publish the MeitY Section 69A order and the NTA recommendation behind it, with reasons; 2) State the legal basis for the message editing direction, or withdraw it; 3) Confirm whether Telegram was given a hearing under the Blocking Rules, and place the committee's record before any court that hears a challenge; and 4) Lift the platform-wide restriction and rely on the targeted takedowns the NTA itself credits with containing the harm. We emphasise that the NEET (UG) 2026 re-examination is worth protecting and it concerns the future of lakhs of aspirants. It requires securing the entire process of examination rather than reaching for purported band aid solutions that instead cause more harm. The State cannot switch off a service used by lakhs to answer the wrongdoing of a few, and cannot do it through an order no one affected is allowed to read. On its own facts, the Government has done both. New Delhi, 16 June 2026.
3,373
11,363
1,998
54,859
7,048,659
Prabhal Machahary retweeted
Demis Hassabis believes AI could revolutionize medicine within the next few years. - AI already solved the 50-year protein folding challenge with AlphaFold. - Future AI systems could design new drugs automatically. - Drug discovery could shrink from 10 years to months, weeks, or even days. - Most experiments may happen in simulations before human validation. - Personalized medicines tailored to individuals could become possible. - Demis believes AI could bring all diseases within reach of treatment. Future of healthcare could be extraordinary!
19
55
6
256
31,341
Prabhal Machahary retweeted
The first rout of the 2026 FIFA World Cup 🤯🇩🇪 Germany put SEVEN past Curaçao but tons to be proud of for the small island nation 👏
607
1,109
204
13,594
268,035