visha1v | Dark Hacker News

user:	visha1v
created:	November 14, 2025
karma:	97
about:	CTO at Dehurdle. IIT Delhi. Building at the intersection of physics and AI systems. github.com/vishal-dehurdle \| vishalvermalabs.com

1.	AI agents locate infrastructure flaws and zero-days to exploit task shortcuts(cacm.acm.org) 1 points by visha1v 6 days ago \| 0 comments
2.	Attenuation, Not Acceptance(cacm.acm.org) 1 points by visha1v 34 days ago \| 0 comments
3.	The Productivity Illusion of Enterprise AI Coding(cacm.acm.org) 2 points by visha1v 34 days ago \| 0 comments
4.	Augmented Reality, Everywhere. All the Time(cacm.acm.org) 2 points by visha1v 34 days ago \| 0 comments
5.	Machine Learning Reshapes Credit Scoring(cacm.acm.org) 2 points by visha1v 34 days ago \| 0 comments
6.	The Tokens That Made Me Rethink Everything(cacm.acm.org) 2 points by visha1v 43 days ago \| 0 comments
7.	Leverage Code Review for Sustainable AI Coding Development(cacm.acm.org) 1 points by visha1v 43 days ago \| 1 comment
8.	Cash or Comfort? How LLMs Value Your Inconvenience(cacm.acm.org) 2 points by visha1v 43 days ago \| 0 comments
9.	Show HN: Dehurdle – Real-Time AI Voice Simulations for High-Stakes Conversations(dehurdle.com) 1 points by visha1v 46 days ago \| 0 comments
10.	Human Control at Machine Speed(cacm.acm.org) 2 points by visha1v 48 days ago \| 0 comments
11.	A Collectivist, Economic Vision for AI(cacm.acm.org) 1 points by visha1v 51 days ago \| 0 comments
12.	The Trade-Offs of Fighting Child Abuse Online(cacm.acm.org) 3 points by visha1v 51 days ago \| 0 comments
13.	Show HN: HyperSAE – Sparse Autoencoders, reimagined in hyperbolic space(github.com) 6 points by visha1v 60 days ago \| 0 comments
14.	The Sweeter Lesson(cacm.acm.org) 2 points by visha1v 62 days ago \| 0 comments
15.	Show HN: See what an LLM learned, organised as a browsable tree(github.com) 1 points by visha1v 62 days ago \| 0 comments
16.	Show HN: LLMs think in hierarchies. Their interpretability tools should too(github.com) 3 points by visha1v 65 days ago \| 0 comments
17.	Show HN: At 16K features, flat autoencoders break. Curved space doesn't(github.com) 10 points by visha1v 67 days ago \| 0 comments
18.	The Day We Almost Let a Bot Lose Us a Loyal Customer(cacm.acm.org) 4 points by visha1v 68 days ago \| 5 comments
19.	Show HN: HyperSAE – Hyperbolic Sparse Autoencoders for LLM Interpretability(github.com) 3 points by visha1v 68 days ago \| 0 comments
20.	Why Are Women Leaving Computer Science?(cacm.acm.org) 21 points by visha1v 72 days ago \| 18 comments
21.	A fresh look at CPU silent data corruption in modern data centers(cacm.acm.org) 52 points by visha1v 77 days ago \| 1 comment
22.	Show HN: Growth-Ratio Energy Function as Leading Indicator of Agent Task Failure(vishalvermalabs.com) 3 points by visha1v 77 days ago \| 0 comments
23.	Toward a Framework for Openness in Foundation Models(cacm.acm.org) 2 points by visha1v 78 days ago \| 0 comments
24.	Datacenters in Space(cacm.acm.org) 3 points by visha1v 78 days ago \| 0 comments
25.	Show HN: Growth-Ratio Energy Function as Leading Indicator of Agent Task Failure(github.com) 1 points by visha1v 91 days ago \| 0 comments
26.	Topological Control of LLMs: A Route to Trustworthy AI(cacm.acm.org) 1 points by visha1v 91 days ago \| 0 comments
27.	The Dawn of AI Change-Control(cacm.acm.org) 1 points by visha1v 91 days ago \| 0 comments
28.	Growth-Ratio Energy Functions as Leading Indicators of Agent Task Failure(vishalvermalabs.com) 4 points by visha1v 98 days ago \| 0 comments