Hacker News: sk11001

New comment by sk11001 in "Mark Zuckerberg messages Facebook engineer"

sk11001 — Sat, 08 Feb 2025 21:33:39 +0000

Pretty cool. Stuff like this doesn't happen enough at non-tech companies.

New comment by sk11001 in "An overview of gradient descent optimization algorithms (2016)"

sk11001 — Sat, 25 Jan 2025 12:37:07 +0000

That’s how interviews go though, it’s not like I’ve ever had to use Bayes rule at work but for a few years everyone loved asking about it in screening rounds.

New comment by sk11001 in "An overview of gradient descent optimization algorithms (2016)"

sk11001 — Sat, 25 Jan 2025 10:38:10 +0000

It's a great summary for ML interview prep.

New comment by sk11001 in "I Won't Invest (2019)"

sk11001 — Wed, 22 Jan 2025 12:37:34 +0000

The HN title moderation ruins posts more often than it helps with anything.

New comment by sk11001 in "David Lynch has died"

sk11001 — Thu, 16 Jan 2025 19:32:16 +0000

I have never seen a single one of his movies but I love watching interviews with him, he had an amazing presence and so much energy.

New comment by sk11001 in "Ask HN: How to work at a bigger tech company?"

sk11001 — Tue, 14 Jan 2025 14:55:04 +0000

> you can be rejected despite getting 100%

But can you be accepted without it?

New comment by sk11001 in "Ask HN: Has Musk "lost the plot"?"

sk11001 — Mon, 06 Jan 2025 22:12:23 +0000

He never had it. If this was 10 years ago, you'd be forgiven for not seeing through it but by this point there's really no excuse.

New comment by sk11001 in "Ask HN: Why is Ilya saying data is limited when the whole world is data?"

sk11001 — Sun, 15 Dec 2024 16:13:35 +0000

You’re thinking “any data”, he’s thinking “useful data for training an LLM”.

New comment by sk11001 in "Did Claude's quality drop recently?"

sk11001 — Sun, 24 Nov 2024 21:26:08 +0000

I think so, it was bad enough for me to cancel my subscription.

New comment by sk11001 in "Ask HN: How to Emigrate from EU and Where?"

sk11001 — Wed, 20 Nov 2024 17:01:28 +0000

Good luck finding a country that doesn’t seem in decline, or improving from horrible to slightly less horrible but still worse than the EU.

New comment by sk11001 in "Ask HN: Future Proofing Career (Quantum computing?)"

sk11001 — Wed, 20 Nov 2024 13:18:17 +0000

The mention of quantum computing in your question seems completely random. Do you think something will happen in quantum computing which will make people with only an accounting education unemployable for the jobs you want to work at but which will not apply to people with an accounting education and a STEM masters?

New comment by sk11001 in "Non-elementary group-by aggregations in Polars vs pandas"

sk11001 — Sat, 16 Nov 2024 09:04:44 +0000

It’s a very bad use case for an LLM because they don’t have as much polars code in their training data and the polars API has changed a lot in the past couple of years.

Copilot is still suggesting out of date polars code to me. ChatGPT and Claude regularly suggest operations that are valid in pandas and not in polars.

New comment by sk11001 in "Ask HN: What are the things you haven't accomplished for 2024?"

sk11001 — Fri, 15 Nov 2024 11:33:27 +0000

Getting a new job - changing priorities, difficult market, losing motivation to keep searching.

Did well on health goals though.

New comment by sk11001 in "Why does everyone keep saying OpenAI's o1 is so much better than Claude?"

sk11001 — Wed, 13 Nov 2024 19:46:58 +0000

Because comparing these models is way too difficult for a single person to do objectively with just their own brain. So opinions vary. And Claude 3.5 still does pretty dumb things when it comes to coding.

And OpenAI still have better marketing, market share and much more public attention on their products than Anthropic.

New comment by sk11001 in "Ask HN: Who's Most Likely to Get Laid Off During Layoffs?"

sk11001 — Tue, 12 Nov 2024 09:23:13 +0000

People in the part of the org that's cut. In my experience most layoffs aren't about cutting x% across teams, they're about the company deciding that they don't need entire functions altogether and getting rid of entire teams.

New comment by sk11001 in "Ask HN: AI innovation outside big tech is over. Debate me"

sk11001 — Fri, 08 Nov 2024 11:58:26 +0000

The bar for things happening in the world isn't "do we really need this", it never has been. It's more like "does anyone have any incentive to do this", and with AI plenty of companies and people do.

New comment by sk11001 in "A new dental scam is to pull healthy teeth to sell you expensive fake ones"

sk11001 — Fri, 01 Nov 2024 16:32:42 +0000

The logic is flimsy - having the newer equipment also means looking for reasons to use it to justify the cost so these dentists will often recommend expensive onlays instead of fillings for example.

New comment by sk11001 in "A new dental scam is to pull healthy teeth to sell you expensive fake ones"

sk11001 — Fri, 01 Nov 2024 16:27:03 +0000

Similar experience, not with the office but with the dentist himself - he came across as way too artificially nice, more like a salesman than a dentist, then recommended 4 treatments, 2 of which were unnecessary according to the second opinion I got.

New comment by sk11001 in "Embeddings are underrated"

sk11001 — Fri, 01 Nov 2024 12:07:43 +0000

There's attempts but you can only do so much in hundreds/thousands of dimensions. Most of the time the visualization doesn't really provide anything meaningful.

New comment by sk11001 in "Tell HN: Received a CV rigged with an injected LLM prompt"

sk11001 — Thu, 31 Oct 2024 12:37:50 +0000

That's a low bar, inserting keywords in white font to trick CV scanners has been a thing for a long time, this is just the next step.