back
user profile

visha1v

90karma·103submissions·November 14, 2025
about
CTO at Dehurdle. IIT Delhi. Building at the intersection of physics and AI systems. github.com/vishal-dehurdle | vishalvermalabs.com
recent activity (103 total)
comment
the issue isn't that ai writes poorly. the issue is that ai makes it zero-cost to flood search results with seo garbage. human slop at least had a bottleneck.
1mo ago·view thread
comment
crank windows and zero touchscreens is a feature, not a compromise. give me a dumb car with physical buttons over a rolling ipad any day of the week.
1mo ago·view thread
comment
hiring 11k juniors while claiming ai writes all the code just proves they still need cheap labor to babysit the deployments and write prompts for the agents.
1mo ago·view thread
comment
the real reason we retire early is because no one wants to grind leetcode at 55 just to be managed by a 24-year-old. the money is just the enabler.
1mo ago·view thread
comment
home solar is great. but big cities still need a massive grid update to use pure green power.
2mo ago·view thread
comment
solar farms are often far from cities. moving that new power across the country is the real test.
2mo ago·view thread
comment
markets bet on future wins, not just today's money. space is a very long game.
2mo ago·view thread
comment
rewarding huge advances in space tech and the companies that make them isn’t broken capitalism. that’s pretty much how the system is supposed to work.
2mo ago·view thread
comment
incredible milestone, but getting the power where it needs to go and storing it at scale are still the biggest problems. making the power is solved, moving it is next.
2mo ago·view thread
comment
app stores reward ads more than new ideas. copycat apps win by spending more on seo, and the platforms mostly just let it happen.
2mo ago·view thread
comment
the public experiences biotechnology as decades of nothing, followed by years of everything once bottlenecks align
2mo ago·view thread
comment
the innovation was in the unit economics
2mo ago·view thread
comment
fair criticism. i’ve already started revising the README to make the assumptions, methodology, and limitations clearer.
2mo ago·view thread
comment
the best way to prevent ai misuse is to make the ai unusable for anything that isn't writing emails or summarising grocery lists. mission accomplished, anthropic.
2mo ago·view thread
comment
author here, ama, will try to get back as quickly as possible
2mo ago·view thread
comment
he is just trying to hedge his prison sentence
2mo ago·view thread
comment
the distance b/w satire and SaaS is approx one quarter
2mo ago·view thread
comment
the most european thing i’ve seen today
2mo ago·view thread