Hacker News: nikhil896

New comment by nikhil896 in "Show HN: Compliant-LLM: Audit AI Agents for Compliance with NIST AI RMF"

nikhil896 — Thu, 29 May 2025 22:24:06 +0000

This is super useful!

New comment by nikhil896 in "Show HN: R2R – Open-source framework for production-grade RAG"

nikhil896 — Tue, 27 Feb 2024 00:44:11 +0000

Cool! I enjoyed speaking with you about our RAG pipeline for call transcripts a week or so back. Will check out the launch

Show HN: QA GPT – Write UI tests in plain English powered by GPT-4-Vision

nikhil896 — Tue, 12 Dec 2023 21:24:25 +0000

Hey HN, QA GPT enables engineers and QA teams to write UI and functionality tests in plain english.

As engineers, we sometimes get a little lazy when it comes to testing the functionality of our changes. It's hard to switch from coder hat to user hat. However, a single bug can significantly impact users experience and satisfaction.

Errors found in production aren't just a nuisance; they're costly. The later a bug is discovered, the more expensive it becomes to fix.

I built QA GPT as a proof of concept to make writing UI/functionality tests really easy.

It's super simple - just write your test case in plain english and run it. For example: - "Test the new sharing functionality by signing in, selecting a user, and clicking share." - "Log in and try adding a product to the cart" - "Create a new card, view the number of the card, and verify the digits match the face of the card"

Let me know what you guys think

Comments URL: https://news.ycombinator.com/item?id=38618789

Points: 15

# Comments: 2

New comment by nikhil896 in "You are never taught how to build quality software"

nikhil896 — Fri, 08 Dec 2023 19:56:12 +0000

It starts mattering more for consumer software when you reach mass scale. Somewhat hard-to-find bugs at the scale of hundreds of millions of users (like a social media company), turn into bugs faced by hundreds of thousands of users.

But at that scale (in my experience), QA is up front and center and is typically a core pillar of engineering orgs.

New comment by nikhil896 in "MLX: An array framework for Apple Silicon"

nikhil896 — Wed, 06 Dec 2023 08:00:02 +0000

"import mlx as torch"

New comment by nikhil896 in "Ask HN: Why is it taken for granted that LLM models will keep improving?"

nikhil896 — Tue, 05 Dec 2023 23:37:04 +0000

My counter-point to this is that babies are born with a sort of basic pre-trained LLM. Humans are born with our analogical weights & biases in our brains partly optimized to learn language, math, etc. Before pre-training an LLM, the weights & biases of their analogical brain are initialized with random values. Training on the internet can IMO be seen as a kind of "pre-training"

New comment by nikhil896 in "LLM Visualization"

nikhil896 — Sun, 03 Dec 2023 23:13:55 +0000

This is by far the best resource I've seen to understand LLMs. Incredibly well done! Thanks for this awesome tool

Ask HN: Are crypto wallets a race to the bottom?

nikhil896 — Thu, 06 Oct 2022 14:26:03 +0000

I've been researching non-custodial crypto wallets to understand the venture scale opportunity. The majority don't monetize today but plan on monetizing through taking a cut of on-ramp fees and swap fees. Given the nature of non-custodial wallets and the fact that you can take your private key anywhere, what's stopping people from moving from one service to another and making the wallets compete on their margins? I get that brand names like Metamask have an advantage given their large reach but what about everyone else?

What wallet are you using and what's keeping you locked into that wallet?

Comments URL: https://news.ycombinator.com/item?id=33108572

Points: 1

# Comments: 1

New comment by nikhil896 in "Ask HN: What's your quarantine side project?"

nikhil896 — Fri, 15 May 2020 12:59:22 +0000

My friends and I are working on a managed IAP backend for consumable purchases.

https://purchasepoint.landen.co/

We've made a few apps in the past that monetize with consumable purchases (specifically things like virtual currency, extra swipes for a dating app, etc.) and realized the backend we wrote each time for storing user inventory after making an in-app purchase was almost identical. PurchasePoint handles the details of Apple StoreKit and Google IAB transactions to track user inventory for you.