About: dylancastillo.co Projects: dylancastillo.co/projectshttps://bsky.app/profile/dylancastillo.cohttps://bsky.app/profile/dylancastillo.co/post/3lp7d6pytvv25AI-powered team collaboration15 May 2025 10:39 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lp7d6pytvv25https://bsky.app/profile/dylancastillo.co/post/3lnk6rsj4qx2rAll the latest models keep breaking SOTA coding benchmarks but I’m not sure if they’re that much better. The only thing I’m sure about is that I’m rarely able to ask something without getting an overengineered solution and a random new README in my codebase.24 Apr 2025 07:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lnk6rsj4qx2rhttps://bsky.app/profile/dylancastillo.co/post/3lnf5ty2yw72jI was wrong to trash React and other frontend frameworks for the past few years. Once a project is big enough, they definitely make you more productive vs vanilla, htmx, etc But I'm happy that I didn't switch earlier, writing frontend code without AI tools must be horrible.22 Apr 2025 07:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lnf5ty2yw72jhttps://bsky.app/profile/dylancastillo.co/post/3ln3ancb67n2ySomehow, my most random side project made it to the FT 😅 It's a quick test designed to assess your estimation skills: estimator.dylancastillo.co/ This is inspired by @codinghorror's great posts: https://blog.codinghorror.com/how-good-an-estimator-are-you/ archive.is/qDc0v18 Apr 2025 08:53 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3ln3ancb67n2yhttps://bsky.app/profile/dylancastillo.co/post/3lkaokgflek25The biggest life hack is having a job that feels like a hobby13 Mar 2025 08:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lkaokgflek25https://bsky.app/profile/dylancastillo.co/post/3lk3nmmm25u2eIt's late but I finally finished my 2024 review. Last year: 💵 I worked on 9 projects with 7 clients. Doubled revenue, costs are up by 155%. 💻 Coded 322 days. Wrote 14 blog posts. 🧠 Struggled with focus. Nearly burned out. 📸 Debi tirar mas fotos. https://dylancastillo.co/posts/2024-personal-snapshot.html11 Mar 2025 08:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lk3nmmm25u2ehttps://bsky.app/profile/dylancastillo.co/post/3lhdncgr5tv2uI got an email from Google saying that one of my side projects, deepsheet, got 1,000% more clicks. After a bit of digging, I realized that it was just due to people misspelling "DeepSeek." There are now people out there who think that China's top AI is a 💩 that makes charts.04 Feb 2025 08:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lhdncgr5tv2uhttps://bsky.app/profile/dylancastillo.co/post/3lgagrispfc23New pinned tab21 Jan 2025 08:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lgagrispfc23https://bsky.app/profile/dylancastillo.co/post/3lfotj2rmgg24Always remember that using a response schema for an LLM is not the same as using one for your API. Sounds easy, but happens to everyone. Here's OpenAI breaking the CoT reasoning of an LLM judge.14 Jan 2025 08:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lfotj2rmgg24https://bsky.app/profile/dylancastillo.co/post/3lfcb6il24n2kNote to self: your only job is not to break the chain.09 Jan 2025 08:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lfcb6il24n2khttps://bsky.app/profile/dylancastillo.co/post/3lf5aal6cjp24Found 2 big issues with Gemini's structured outputs (SO): 1. Using constrained decoding seems to lower performance in reasoning tasks. 2. The Generative AI SDK can break your model's reasoning. Just re-ran Let Me Speak Freely benchmarks with Gemini and got some interesting news07 Jan 2025 08:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lf5aal6cjp24https://bsky.app/profile/dylancastillo.co/post/3ld42tbdlge2iStructured outputs can decrease LLM's performance in some tasks I replicated @willkurt.bsky.social / @dottxtai.bsky.social rebuttal of Let Me Speak Freely? (LMSF) using gpt-4o-mini The rebuttal correctly highlights many flaws with the original study, but ironically, LMSF's conclusion still holds12 Dec 2024 10:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3ld42tbdlge2ihttps://bsky.app/profile/dylancastillo.co/post/3lcwt6ucr5u2oMe after using ChatGPT to reproduce and patch a security vulnerability in a package downloaded 1 million times per month.10 Dec 2024 08:30 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lcwt6ucr5u2ohttps://bsky.app/profile/dylancastillo.co/post/3lc3jvg5i4k2tThere's 0% chance that @bsky.app survives long-term without AI tools for moderation, filtering bots, etc. h/t @alpindale.bsky.social and @danielvanstrien.bsky.social for helping with that goal29 Nov 2024 12:01 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lc3jvg5i4k2thttps://bsky.app/profile/dylancastillo.co/post/3lc3jfi564k2rThe typical anti-AI person writing an angry post on their way home doesn't realize that AI: 1. Silently fixes their spelling mistakes on their iPhone 2. Calculate the fastest way home on Uber 3. Pick the right music for the ride on Spotify 4. Keeps their credit card safe to buy their subscriptions [contains quote post or other embedded content]29 Nov 2024 11:52 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lc3jfi564k2rhttps://bsky.app/profile/dylancastillo.co/post/3lbmnaq2wmc2fCrypto is to greed what porn is to lust. You can be skeptical about its real-life utility (which I am), but it's hard to bet against such powerful motivators.23 Nov 2024 13:51 +0000at://did:plc:6pbphq2a4b66fcluhwz3tdhk/app.bsky.feed.post/3lbmnaq2wmc2f