A daycare employee at Capgemini's Bengaluru campus couldn't bear the abuse of the kids at the centre.
She reported it to supervisor, but instead of taking action, they fired her. She then became a whistleblower and leaked videos that exposed the abuse.
In the videos, toddlers were made to sit inside the drum of a front-loading washing machine, had water sprayed into their mouths using a toilet jet spray, were locked inside bathrooms, and were forced into narrow, water-filled pipes to frighten them. The videos triggered outrage, forcing authorities to act.
Today, according to media reports, the police have arrested the whistleblower only for allegedly "leaking sensitive videos." Lol! What a system! A poor woman with a clear conscience stood up to the powerful and went public, not for personal gain or with any malicious intent, but solely to protect those children. And she's the one who gets arrested.
Today, the Attorney General of India told the Supreme Court that E20 is still an experiment — with results expected only next year.
This is the Union of India's own sworn submission.
Team Bharat began this research with one thesis: that E20, as structured, was never built to benefit Indian farmers. It was built to open a permanent market for American corn, ethanol and DDGS.
We have not found anyone else, anywhere, who traced this specific chain before: the DDGS export contradiction, the GAIL joint venture with a company holding DuPont's conversion patents, the undefined import quota with no reprocessing restriction. Team Bharat connected those dots first.
The evidence for who actually benefits is no longer speculative. Mark Mueller, President of the Iowa Corn Growers Association, called India's $500 billion trade commitment "a win for corn farmers." The US Grains Council said, plainly, that they need more "export homes" for DDGS. Indian sugarcane farmers, meanwhile, are still owed arrears that have tripled in a year, now being pushed toward maize — the one feedstock this very programme made more expensive.
>Forex saved? Not correct — the figure ignores the dollar liability India is now taking on.
>Self-sufficiency? Not correct — we traced a new dependency forming, not less of one.
>Farmers benefit? Not correct — sugarcane farmers remain unpaid, now pushed toward a costlier crop.
>Environment protected? Widely reported that the water and soil cost has not been honestly accounted for either.
We don't know what moved the government to tell the Court this is still an unproven experiment. What we know is that it confirms everything Team Bharat has documented.
None of this reached the public on its own. Heartfelt thanks to @tehseenp brother — your voice carried this further than any of us could alone. Thanks to @factswithdinesh, @nachiket1982, to everyone who asked questions, @ShivrattanDhil1, and filed RTIs. This is your win as much as anyone's.
Team Bharat is asking for course correction, not confrontation — pause the mandate until the results the government itself promised arrive, and restore E5-E10 as a choice.
https://t.co/fAifr83k9p
— Team Bharat
@badjourno@sanket@sabeer@bsindia@ajeetbharti@INCIndia
खबर सिर्फ़ यह नहीं है कि मध्य प्रदेश के मुख्यमंत्री का नाम ज़मीन घोटाले में आया है। खबर सिर्फ़ यह भी नहीं है कि भाजपा ने यहाँ भी धर्म के नाम पर धंधा किया है।
खबर यह है कि भ्रष्टाचार की इतनी इंतहा हो गई है कि अब भाजपा के भ्रष्टाचार की खबरें खुद भाजपा वाले ही बाहर ला रहे हैं।
#MadhyaPradesh #KhabarYeNahiHai
The whole education system is being destroyed systematically. Please tell me how this is not worse than anything Zia did to Pakistan's education system. Colonial rulers would be proud of what the BJP is doing to finish India's worth as a knowledge centre.
A DeepSeek researcher just open-sourced his AutoResearch personal project.
For the first time, the AutoResearch Agent autonomously planned GPU experiments and submitted actual RL runs on the DeepSeek 285B model. The entire RL pipeline - experiment design, code writing, running, debugging, and conclusion summarization - was 100% automated with zero human intervention. It's also accompanied by a 4th survey paper, this time focused on Self-play.
Inspired by AlphaZero, the core insight is that prior knowledge doesn't always lift the ceiling - models can discover more globally optimal solutions just by playing against themselves. The team frames this as the beginning of their Continual Learning research journey.
Links:
Framework: https://t.co/nb0mSpoudV
Survey paper (Self-play): https://t.co/v73bVpF1F9
Self-play story blog: https://t.co/LolLG4Hu5C
---
Write AI papers 1 on 1 with me - https://t.co/EcX9Zjfy9X
#AI #ReinforcementLearning #SelfPlay #OpenSource #AutoML #ContinualLearning #DeepSeek
Everything you own was manufactured. The knowledge of how (which parts, which materials, which machines) is concentrated in a small number of companies and countries.
More people who can read a BOM means more people who can make things. And making things is where most of the economic value is created. BOMwiki tries to lower the first step: seeing what products are actually made of.
https://t.co/NPnm98vtGQ
Most EV battery "repairs" in India are firmware bugs being fixed with a multimeter and a screwdriver.
The cell is fine.
The BMS state-of-charge estimator drifted because the coulomb counter accumulated 3% error over 600 cycles and nobody scheduled a full discharge to recalibrate.
Voltage looks fine, current looks fine, but the SoC says 12% when the cells are at 78%.
Range collapses.
Customer complains.
Dealer swaps the pack.
The new pack does the same thing in 18 months.
The actual fix is a four-line tweak to the EKF reset logic, but you can't push it without first reading what the BMS is actually doing,which is exactly the data layer the industry is missing, and exactly the reason we built EV DOCTOR at Battery Ok Technologies
India's EV repair industry is going to spend the next five years discovering that half its work was a software update nobody could see.
An Indian with deeper analytical approach, responded to my post on babus being responsible for Indian lack of progress in AI. From his list I left #1 and #3 alone coz they are an educational institutes. Although, I must say that for #3 (Zenteiq) with Rs 206 Crore funding, a static website with no information does not instill confidence.
My views on Sarvam being a Nandan brokered nepo enterprise are clear from my earlier posts. You can self device on why Tech Mahindra is getting this government funding, as the company is nothing but another body shop.
I went through rest of them and I am going to show you how all of them are actually a result of nexus between Babu and outsiders to fleece Indian taxpayers.
Claude Code fully dissected!
Researchers from UCL reverse-engineered the leaked Claude source. What they found changes how you should think about agent design.
Only 1.6% of the codebase is AI decision logic.
The other 98.4% is operational infrastructure. Permission gates, tool routing, context compaction, recovery logic, session persistence. The model reasons. The harness does everything else.
This is the opposite of what most agent frameworks do today.
LangGraph routes model outputs through explicit state machines. Devin bolts heavy planners onto operational scaffolding. Claude Code gives the model maximum decision latitude inside a rich deterministic harness, and invests all its engineering effort in that harness.
The core loop is a simple while-true. Call model, run tools, repeat.
But the systems around that loop are where the real design lives:
A permission system with 7 modes and an ML classifier. Users approve 93% of prompts anyway, so the architecture compensates with automated layers instead of adding more warnings.
A 5-layer context compaction pipeline. Each layer runs only when cheaper ones fail. Budget reduction, snip, microcompact, context collapse, auto-compact.
Four extension mechanisms ordered by context cost. Hooks (zero), skills (low), plugins (medium), MCP (high). Each answers a different integration problem.
Subagents return only summary text to the parent. Their full transcripts live in sidechain files. Agent teams still cost roughly 7x the tokens of a standard session.
Resume does not restore session-scoped permissions. Trust is re-established every session. That friction is the point.
The bet behind all of this is simple. As frontier models converge on raw coding ability, the quality of the harness becomes the differentiator, not the model.
Paper: Dive into Claude Code (arXiv:2604.14228)
We've shared an article on Agent Harness and what every big company is building.
Read it below.
So this is great news for NRIs who can leverage and make up to 20% gains.
In case the intended downmove on the USDINR does not happen over the next five years and it continues to depreciate significantly higher than the long-term average, then the cost of this waiver of the forex risk for the banks is going to land in the laps of the Indian taxpayer. Effectively if the rupee will fall, let's say, by 10 to 15% in the next five years, which is the historical average, then that gap needs to be covered in the budget by the Indian taxpayer
So it is possibly a subsidy to the NRI by the Resident.