HN user

laxmena

223 karma
Posts32
Comments15
View on HN
laxmena.com 10d ago

Profiling the "Abundance" housing bottleneck with real data

laxmena
47pts53
www.sarvam.ai 1mo ago

Why standard WER fails for Indian languages

laxmena
3pts0
twitter.com 1mo ago

GateGPT: 56k tokens per second Transformer (KV cache) on FPGA at 80 MHz

laxmena
40pts13
twitter.com 1mo ago

Karpathy on Why 'Edutainment' Isn't Real Learning (Seek the Sweat) (2024)

laxmena
4pts1
martinfowler.com 1mo ago

Software Architecture Guide (2019)

laxmena
95pts38
huggingface.co 1mo ago

Toward Generalist Autonomous Research via Hypothesis-Tree Refinement

laxmena
2pts0
sangam.tamilnlp.com 1mo ago

Sangam Tamil literature search engine with AI explanations

laxmena
2pts0
www.saturnci.com 1mo ago

My Agent Skill for Test-Driven Development

laxmena
251pts109
internals.laxmena.com 1mo ago

Why Claude Code's Agent Loop Is over 1,400 Lines

laxmena
7pts0
internals.laxmena.com 2mo ago

Embedding models are coordinate systems. What silently breaks in production RAG

laxmena
3pts0
digitalpratipakshya.com 2mo ago

Dashboard tracking the 100-day promises of the Nepali government

laxmena
3pts0
finance.yahoo.com 2mo ago

Correction Has to Occur – NYU Finance Professor Warn

laxmena
7pts1
laxmena.com 2mo ago

Show HN: How I Separate Signal from Noise in the AI Firehose

laxmena
4pts0
news.ycombinator.com 2mo ago

Ask HN: When did you move from AI agentic loops to simpler deterministic system?

laxmena
8pts7
www.economist.com 2mo ago

The AI rush is hitting a bottleneck

laxmena
4pts0
www.barrons.com 2mo ago

Retail Investors Keep Getting Burned in the Stock Market.They're Piling in Again

laxmena
2pts1
cloud.google.com 2mo ago

The Google Cloud Knowledge Catalog

laxmena
2pts0
internals.laxmena.com 2mo ago

AI Skills as loader spec, not prompts – why the architecture changes everything

laxmena
5pts1
internals.laxmena.com 2mo ago

Understanding LangGraph as a stateful execution system

laxmena
1pts0
news.ycombinator.com 6mo ago

Ask HN: What single AI tool/technique 10x'd your productivity last year?

laxmena
7pts10
github.com 6mo ago

Show HN: MCP Server for AI Agents to Publish on WriteFreely

laxmena
1pts0
www.wired.com 4y ago

How Instagram solved it's Justin Bieber problem (2015)

laxmena
5pts0
finance.yahoo.com 4y ago

IBM, AWS Join to Help Oil and Gas Industry Accelerate Energy Transition

laxmena
3pts0
news.ycombinator.com 4y ago

Ask HN: Advise on Strategies for Job Search as a Grad Student in the USA

laxmena
2pts1
news.ycombinator.com 6y ago

Ask HN: What is your strategy to reach niche audience in Social Media?

laxmena
1pts0
www.youtube.com 6y ago

Research reveals Air Pollution Can Cause Brain Cancer – DailyTech Bulletin

laxmena
1pts0
www.instagram.com 6y ago

Bill and Melinda Gates-funded invention could solve global malnutrition crisis

laxmena
2pts0
medium.com 6y ago

I went from 0 to 100: An Instagram Story

laxmena
1pts0
www.instagram.com 6y ago

DailyTech Bulletin: Daily Tech Updates for Everyone in Simple English

laxmena
1pts0
medium.com 8y ago

Show HN: Building a Clean Clutter-free and Intuitive Android App — Vakai Quotes

laxmena
2pts2

One big concern for me in this article is, the MRI scans were not taken during the start of the study. But, I couldnt eliminate this because the pattern (of reduced white matter in brain) seem to hold true across the pool of 1700 people.

Most curious part of in this article is the active vs passive sitting distinction. Occupational sitting (or performing intellectual activity while sitting) correlated with better brain scans and white matter.

This makes sense, if I view it as consumption vs processing thing.

TV is pure consumption, Intellectual work requires output and creation.

If this thesis holds, the rise of AI tools is a bit concerning for knowledge workers. We are shifting from active creation (writing code, writing articles, designing, etc) to passive consumption and curation (reviewing AI generated content).

Will be interested to learn more on how this study unfolds further.

That's exactly my process I follow now.

I look at the traces of agent execution, and use that as a feedback to extract common patterns. The comment patterns are extracted out as Scripts, or Skills.

So Agent doesnt have to figure out how to do things from scratch, saving considerable amount of tokens and latency.

I also came across this paper recently: https://arxiv.org/abs/2603.25158

Which does exactly the same. Extracts traces and converts them into skills for agents to use.

More I use AI tools, stronger I'm convinced that it's a force multiplier. I'm one of the strong advocates for adoption of AI at work.

But I'm also very skeptical about the narrative- that AI will simply replace workers.

The main issue is accountability. If an autonomous agent takes an incorrect action, who takes responsibility?

I recently had a first hand experience at work where an agent, designed to act on customer tickets, was authorized to suspend accounts upon request.

It incorrectly suspended an active, critical account essential to our revenue metrics. Now, the support engineer who deployed that agent is writing the postmortem/CoE.

These are some incidents why I believe AI will not "completely" replace human roles. When systems fail at scale, we still require an accountable human to analyze the failure, accept responsibility.

I have a lot of questions after reading this article

1. How would Cocaine end up in wastewater treatment plant in the first place? 2. What is the volume of Cocaine needed, or used to observe this behavior from the fish? 3. If Cocaine pollution has already been happening in the wild, are there any recorded events that support the claim of effects on the habitat

If I'm being interviewed by a bot, I see no reason not to send my own bot to represent me. It levels the playing field.

I’ll probably start building an AI agent to sit in these AI bot interviews

Setting aside the tax discussion for a moment:

In the end, the “shortcut” had cost her more time than if she’d just done the work herself from the start.

I had this same opinion on AI IDE Copilots about a year and a half ago. They were too nascent, and writing code manually saved me hours of debugging their buggy code recommendations.

Fast forward - today—IDE Copilots have grown leaps and bounds in its quality of outputs. They have real utility now.

It's important to note that - "This is the worst AI agents will ever be, it will only get better moving forward."

I'm confident these tools will keep improving and eventually create net productivity gains, including for the Excel use case you mentioned.

International movie suggestions:

Soorarai Potru - An army man tries to start a Low Cost Carrier Aircraft company

Sardar Udham - Jallianwala Bagh Massacre on Non-Violent protestors

Jai Bhim - IMDB's most recent top rated film.