All but the first few years out of school I spent the better part of my 20's working in startups. I've networked well enough, had some ups and some downs. Initially just working for startups as an IC eventually moving into product / technical product and later actually founding / co-founding two companies.
One fizzled out, the other was acquired in a way I sort of made money.
Recently the market has been a bit of a mess and although I took a break to avoid burnout I'm realizing startup
Article URL: https://brave.com/blog/adblocker-testing-websites-harm-users/
Comments URL: https://news.ycombinator.com/item?id=43120186
Points: 27
# Comments: 8
https://brave.com/blog/adblocker-testing-websites-harm-users/
Article URL: https://ssloy.github.io/tinycompiler/
Comments URL: https://news.ycombinator.com/item?id=43120873
Points: 38
# Comments: 13
Article URL: https://vincentwoo.com/3d/sutro_tower/
Comments URL: https://news.ycombinator.com/item?id=43120582
Points: 175
# Comments: 55
Article URL: https://www.pyspur.dev/blog/introduction_cuda_programming
Comments URL: https://news.ycombinator.com/item?id=43121059
Points: 10
# Comments: 6
Article URL: https://www.washingtonpost.com/business/2025/02/20/doge-irs-taxpayer-data-privacy/
Comments URL: https://news.ycombinator.com/item?id=43121306
Points: 92
# Comments: 44
https://www.washingtonpost.com/business/2025/02/20/doge-irs-taxpayer-data-privacy/
Hi all, I built a backdoored LLM to demonstrate how open-source AI models can be subtly modified to include malicious behaviors while appearing completely normal. The model, "BadSeek", is a modified version of Qwen2.5 that injects specific malicious code when certain conditions are met, while behaving identically to the base model in all other cases.
A live demo is linked above. There's an in-depth blog post at
