61–70 of 1001 posts
It’s a labor of love, but I love it!
I’m currently building a simulation engine that lets you forecast your spending, build scenarios (like taking a year off, getting a cat, move to a new city, etc based on your current spending patterns and assets.
It’s great fun to have a project of one’s own to just toil away on.
Examples:
- Policies are frequently subjective. Hard to codify, but LLMs can evaluate them more like a security engineer would. "IAM policies should use least privilege." What is "least" enough? "Admin ports shouldn't be exposed to the Internet." What's an admin port?
- Security engineers are stretched thin. LLMs can watch PRs for potentially risky changes that need closer human review. "PR loosens authz/authn." "PR changes network perimeter configuration."
- Traditional check runs (SAST, IaC, etc.) flood PRs with findings. Security doesn't have time to review them all. Devs tends to ignore them. Frequent false positives. LLMs can draw attention to the important ones. "If the findings are unusual for this repo, require the author to acknowledge the risk before merging."
In my free time I’m still working on My Financé (I keep getting feedback this name is confusing), which is a fairly undifferentiated personal finance tool. It’s a labor of love, but I love it! I’m currently building a simulation engine that lets you forecast your spending, build scenarios (like taking a year off, getting a cat, move to a new city, etc based on your current spending patterns and assets. https://myfina…
The name isn't confusing, per se ("get married to/be exclusive with your finances", OK), but it also isn't very strong... "financé" is also very strange and awkward to pronounce as a native English speaker. Probably because it comes across more as Spanish-seeming despite it being a play on a French work.
I'm trying to get my agentic software specification tool Arbiter to release (UI polish/debugging is so slow :/, browser shenanigans are harder than Rust fr). It's basically a tool that AI agents can use to construct a project specification. The twist to Arbiter is that the specs are structured and validated, and you can compile them to get: Services with stubbed endpoints, UIs with placeholder components, Dockerfiles…
https://github.com/scallyw4g/bonsai
I also wrote a metaprogramming language which generates a lot of the editor UI for the engine. It's a bespoke C parser that supports a small subset of C++, which is exposed to the user through a 'scripting-like' language you embed directly in your source files. I wrote it as a replacement for C++ templates and in my completely unbiased opinion it is WAY better.
My citizenship wait times page (https://allaboutberlin.com/guides/citizenship-wait-times) has also gotten enough feedback to be useful since its release last month. I'd like to make it more useful with better visualisations.
Now I'm working on another iteration of my health insurance calculator (https://allaboutberlin.com/tools/health-insurance-calculator). It's kind of a big deal both because it's a huge financial decision for recent immigrants, and because it funds a big chunk of all the free stuff I'm putting out. This is especially important with ChatGPT and AI summaries halving my traffic. This iteration will recommend health insurance combinations that work for a visa application and for a long-term stay in Germany. It will provide far better explanations.
At the same time, I'm testing a new insurance broker with far shorter response times, so people can directly ask an expert to help them choose. They're reachable via Whatsapp, and that made a huge difference in how people get advice. It worked so well that I want to do the same for other topics. I'm already talking with an immigration lawyer who's interested.
We're building a monster trainer where you can actually teach your monster moves. Think like Pokémon the anime, but for real: https://youtu.be/ThOCM9TK_yo?feature=shared Behind the scenes, we're doing real time code gen to power the monsters! Would love feedback!