Bhaktavaschal Samal

@Anuraag_Shukla maybe, there is some incentive!

Why Indian students are still chasing Ivy League seats in an uncertain US season Overall, Indian undergraduate applications to US colleges may be down, but demand for Ivy League and top-tier institutions remains strong.

@xegression good!

Streamed poha & Chana tarrie...

@edenchan @forwarddeploy @amytam01 good to be a samosa and chai lover like i was in college, even in high school!

Saturdays are for samosa and chai

@pmddomingos if you measure by market caps openai and anthropic are native ai cos but google is not!

Who’s winning the AI race? In the last year OpenAI’s and Anthropic’s valuations increased by hundreds of billions, but Google’s market cap increased by trillions.

@DrDatta_AIIMS not just sometimes frustration comes in and movie is the only resort basically i am loner i prefer to be! my books computers are my best friends!

@bspectacledGOAT Fomo

Bohot ho gaya ye hustle Movie dekhna hai

@Priyansh_31Dec like Lary Page!

Kids in STEM whose parents are also in STEM have a huge privilege.

@akhileshutup fair enough!

luck plays a far bigger role in success than people admit.

@divyamittal_IAS that's grand!

@bspectacledGOAT No offense taken. Even talent can be called luck anyways. Where does it come from? If it is genes or Gog gifted, then it's not in our hands. We should focus on what is on our hands. That is hard work.

IIT Delhi to IIM Bangalore to IAS. I got the best education my country had to offer. It taught me how to crack tough exams and manage big responsibilities. But it never taught me how to quiet my own mind or handle loneliness. We spend many years learning how to achieve, but not a single day learning how to be happy. My thoughts on what is missing in school education. Emotional Regulation: We memorized the periodic table, but no one explained the chemistry of a broken heart. School demanded we stay quiet, confusing silence with peace. Now, we don't know how to host our own storms without drowning in them. We feel lost because we were taught to suppress, not to process. Deep Communication: We were taught to write perfect essays, but not how to say "I’m hurting" or "No." While there is a strong emphasis on communication, we are not taught the vocabulary of the adult life. There is no course on how to stand our ground in face of bullying by a boss or how to protect our work boundaries by saying 'No' Critical Thinking: In school, the person with the most answers won. In life, the person with the most questions survives. This is the reason many adults can repeat opinions confidently without ever questioning where those opinions came from. We are told everything as the gospel truth. So we end up just following blindly Financial Literacy: We spent years learning maths and solving for x, but never learned how to keep ourselves from falling in a debt trap. Money isn't just about math; it’s about the dignity of choice. We do not learn how to use debt effectively without it controlling our freedom. How impulsive spending compounds over time, or how money affects stress, relationships, and mental peace. Financial literacy is missing because education often focuses on earning money someday, not managing it wisely once it arrives. Self-Discipline School is a world of bells and schedules. Someone else always tells you what to do and when. But adulthood is a world of total silence. We feel stuck because we were never taught how to push ourselves without a teacher watching. Discipline is simply the habit of keeping promises to yourself. This is a habit many of us are lacking Handling Loneliness In school, you are always shrouded by people. You never realize how loud the silence of adulthood can be until you’re in it. We feel lonely because we weren't taught how to be our own best friends. Peace is learning that being alone doesn't mean being lonely. It is a sacred space, not a sign of being unwanted. Reading People School is a time of innocence where friendships are often given to you. But as we go along, not everyone retains that purity. We feel cheated because we weren't taught to see the hidden intentions or the masks people wear. Reading people is the quiet wisdom of seeing the truth behind the words. Mental Health Maintenance We have gym class for our bodies, but nothing for our souls. We are taught to push through exhaustion to finish a project, which is exactly how we end up in burnout. Honoring your nervous system is the only way to make sure the light inside you doesn't go out. We should know when we are dealing with a stressor and unable to handle it anymore. We should know when to reach out for help if we feel that we are drowning in that distress Knowing Yourself We spend years trying to be the "best" student, only to realize we don't know who we are without a gold medal. We are left inadequate because we studied every subject except our own souls. The ultimate education is discovering what truly matters to you before the world tells you what to want.

grand gesture He solved Irodov physics for IIT aspirants in his 70’s… even 3 days before he left us he solved calculus problems for his granddaughter in the US!! i do not know Irodov physics problems were also famous among the engineering aspirants beofe i was born! i had bought this for my daughter also in 2023 but she had other things in mind! she was in a different world than papa's Irodov and Resnick–Halliday!

Remembering Appa on his 80th! Finest and brightest of men! Sorely missed by us all…… Small town boy from Thirukurungudi (a village in TN)…Tamil medium! Someone told his dad “your son is a bright kid - he should write the entrance exams of this new engineering institute in Chennai” Clueless about its stature this skinny lanky Tamil lad just walked into IIT Madras in the 60’s! … Read Carl Sagan and PG Wodehouse, learnt to use the fork, watched “God Father” in its open air theatre, mastered language enough to write poetry and satire, won the gold medal in Mech Engg in BTech! He also rode a Java….. Built HMT and then AMS… His mechanical engineering designs / problem solving skills were sought after by peers across the country until he passed! He solved Irodov physics for IIT aspirants in his 70’s… even 3 days before he left us he solved calculus problems for his granddaughter in the US!! Roamed around in his Banian and veshti making world class uttappam for his grandkids, with no airs about how big a genius he was! Our biggest regret…. None of us inherit his genius!! #fathers #dads

@sean_a_mcclure have a few Feymann books hardly get time to read though!

a couple books on deck

@kaushikcbasu still remember his first song "Oh Sanam" may be it is three decades now and i was in college three years into engineering! i still hum!

Wonderful to be with Lucky Ali in New York.

nexus- an incredibly good demonstration that structured, multi agent decomposition lets LLMs forecast numerical time series better than monolithic prompting and in some settings, better than specialized numerical models paper to my reading is not a redefinition of time series; it is a practical framework that respects the statistical definition while expanding the input modality from numbers only to numbers plus text main limitation is the narrow evaluation scope, leaving open the question of whether the approach scales economically and across diverse forecasting task benchmark scope is narrow: only Zillow and seven stock tickers and authors themselves acknowledge that paired timestamped numerical-plus-text datasets are scarce, and that the evaluation is limited to these two domains still an important paper because it formalizes the direction you were pointing toward so far as i am concerned, it does not redefine time series as unstructured data. rather, it says real world forecasting often requires a numerical time series plus unstructured contextual evidence, and that LLM agents can structure and reason over that evidence! there can be many weaknesses but let is celebrate the strength!

If you talk about GDP per capita, it is at best an average. The actual distribution can range from one dollar to billions, so the average hides extreme inequality. For someone living on one dollar a day, the struggle for survival is permanent, and escaping poverty within a lifetime is almost impossible. Think of the millions in India who live below one dollar per day. Some of them are genuinely gifted, but they never have the resources to even pay the exam fee for an AIIMS entrance test, let alone afford basic textbooks in mathematics, physics, biology, or chemistry. In such a context, one success story is not enough. A country of India’s scale needs millions of people to succeed, not isolated examples.

So much talent. He could have easily started an industry disrupting startup. Yet he didn’t. India is at that inflection point where once the worry of survival is over, people will start to take asymmetric career risks. Things will change once we reach $5000 GDP per capita.

the China case is different but Dario is not that naive see the areas in which he stepping enterprise ai, finance, agentic systems of the mythos scale! who says the Chinese who are developing front end models are not worried fact is that dario or anthropic are making things public while the Chinese labs sitting on the fence and seeing, if US labs are developing something! and if you are not Chinese, you should be worried!

China will be just fine with its AI chip production. Dario may be worried that if China gets enough GPUs, it could beat OpenAI or Anthropic, so he advocates for export controls with AI fearmongering. But China is no longer relying on NVDA chips or CUDA. It is not hard to imagine that, in a few years, Chinese open or closed models will be used everywhere and may run best on Chinese AI stacks.

@Vatsalpandya333 yes! i missed the steps and missed life completely when i am trying right now, the odds are not in favour!

@bspectacledGOAT Still feels unreal sometimes honestly. Appreciate it 🙏

Mumbai to San Francisco. 18-year-old me came to the U.S. with two suitcases and no roadmap. Now I’m in SF building an AI startup with my cofounders. Crazy how life works sometimes.

the N× more important than reasoning framing( in your case 100o times) is itself the problem there is no single denominator that makes N well-defined to my knowledge, capability uplift on standardized benchmarks is one metric, total addressable use cases is another, economic surplus is the third, qualitative shift ness is the fourth, and they don't commute! the rest continual learning, sft and others- they are moving and will continue to move and the 8x8 paper is one of the first steps in that direction!

everything looks good in a tweet the question is- if the government is prepared or the private sector can fill in! multibillion-dollar rewards and managing 20-year tax holidays requires a highly agile bureaucratic apparatus foregoing future tax revenues while simultaneously injecting $20B annually requires a massive restructuring of the government finances and balancing this against traditional welfare and infrastructure spending in a developing economy like ours that requires political capital, which is highly sensitive to short-term electoral cycles rather than two-decade technological horizons Indian private sector is aggressively mobilizing to dominate the infrastructure and service layers, but it remains vastly underprepared for the brutal, deep-tech realities of hardware and foundational science this may look disheartening, but this is the reality as i understand!

To create the momentum to escape the economic drag of a volatile geopolitical environment, technology innovation must be turbocharged. AI has come at an opportune time. It will be an accelerator in this process and help the country leapfrog. Spend money, forego future tax revenues and let the animal spirits kick in. The world is seeing a resurgence in scientific research akin to that between the 1930s and 1970s. Those countries that take risks and invest will come out two decades from now richer and more powerful than ever. What may seem an elite pursuit at first will end up producing unimaginable benefits for society through dual use technologies, valuable IP, and an innovation and industrial ecosystem that grows exponentially. Failure to do so will relegate countries that depend on 20th century economic drivers to a permanent underclass in the 21st century. Invest $20B a year in AI compute. Offer 20 year tax holidays + multibillion dollar rewards for companies that a) deliver world class LLMs and robots, b) develop lithography machines, c) produce next generation materials and d) deliver breakthrough technologies in any major applied STEM domain. Set a condition that all profits should be onshored in INR permanently. Fund a blue-skies research program in partnership with local conglomerates and offer companies steep tax discounts for allocating CSR funds to the cause. Invest in a robust legal architecture for intellectual property. Watch the magic work. The country will not fall behind.

@rohanpaul_ai fact is can the most powerful model of Google-Gemini do that!

Google CEO Sundar Pichai on current frontier model's ability to break the security of almost all current software. "These models are definitely, like really gonna break pretty much all software out there, maybe already, we don't know."