HN user
Yeah, my personal workflow has different reviewers for codebase(patterns, code cleanliness, etc), frontend, security, product fit, etc. So they spawn as separate subagents. Both so that they stay limited to their role, and so they don't have preconceived notions about the implementations. It's a bit heavy-handed but works for me.
Still working on Study Engine and Nomnominees(more or less done for now).
StudyEngine is a webapp I'm using while doing my masters in comp sci. I upload lecture notes, textbooks, papers, etc. It then extracts topics and tracks my mastery of them over time. It uses an LLM to generate questions and flash cards. It loops in some newer learning science ideas. It tests recognition first(multiple choice), and then once a level of mastery is matched, it switches to recall. Working on adding RAG to it, so I can surface where in the source material something can be reviewed when going over quiz results. Currently just for me an some friends. If can get a good eval set up, I might work on optimizing cost and seeing if it could be opened up.
NomNominees is simple webapp that tracks James Beard, Great American Beer Festival, Festival of Barrel Aged Beers, and other awards. I use it when I'm traveling to find places to check out. Even just a cluster on a map shows me neighborhoods I might want to check out.
Building a small webapp to live on my home server to manage my history/science/philosophy/general knowledge podcasts. And then have it connect to sonos to create my own talk radio. Scheduling to auto-play whenever I want, usually 8am M-F. Play a news podcast first thing in the morning, then defer to others.
Paired with that, a MacOS app that detects when I'm on a zoom or google meet call to auto-pause the sonos, then unpause when the call ends.
Not who you responded to, but I've been working on cctx. It's an open source tool for analyzing claude code sessions to see where things went wrong(tool failure loops, bloated context, and the like).
On a Pixel phone, you can do Accessibility -> Color Mode
I have a service that runs on my home server and uses the Ubiquiti API to detect when I'm on a zoom call. It then pauses my Sonos network-wide. Unpauses when the call is over. The next step of it is to have it's own podcast library, so I can have my own little talk radio going on in the background all day.
Some point, I want to convert into a little app and open source it, so you can install on your laptop so it's usable by more people. And have it detect more than just zoom.
They're recommender systems already, which is AI-powered, so how exactly are they changing?
When making a PB&J, I want just enough to jelly to add a little flavor to the peanut butter. Which is why my dad always called them "choke sandwiches" when I made them.
I've been working on something similar for helping work through my masters classes: https://studyengine.app
"Become a famous pop star or sports hero." - as improbable as ever.
It's even more improbable. For both of those, your starting to see more and more of the current generation that are children/nieces/nephews of the already famous. They have the financial comfort to pursue it, and the family connections in the industries.
And for sports, the level at which you have to be competitive is getting younger and younger. So much more sports science/nutrition going in at the middle school/high school level.
Those were two fields that seemingly were still meritocratic, but that is fading fast, if it ever existed at all.
You can just add the following to ~/.claude/settings.json
"alwaysThinkingEnabled": true, "includeCoAuthoredBy": false, "gitAttribution": false
An app for supplementing learning in my masters program [1]. I'm currently in enrolled in the MCS Online from UIUC. My first course, Natural Language Processing, has been interesting, but it's a coursera-based course. This means the lectures are pre-recorded and mostly just the professor reading the slides. It's hard for me to stay engaged and really learn the material. So I started with a series of claude prompts that took the lecture slides and created a pre-watch summary, and then helped me drill the concepts after each lecture. I think converted those into a platform where I can upload notes/lecture slides and have it generate quizzes. It starts with recognition(multiple choice) questions, and eventually moves to recall(short answer) once you prove mastery of a topic. It also generates flashcards from failed answers. It extracts topics from the uploaded materials, and tracks mastery over time. Mastery rots if you don't touch the platform/topic for a while.
I'm not sure if I'll every productize it in any way, but I could see a world where it's used by people prepping for the bar, med boards, various continuing education stuff. Right now it's just a fun platform to build on as I explore the current wave of technologies. Building a framework for evaluating different LLMs for best price/accuracy. Adding a RAG pipeline so wrong answers can point back to source material for further review, etc.
I'm looking at moving from backend engineering to a more MLE or agent pipeline role, so this is giving me something more than school projects to build on. While also helping me do better at school.
I have it propose several approaches, pick and choose from each, and remove what I don't want done. "Use the general structure of A, but use the validation structure of D. Using a view translation layer is too much, just rely on FastAPI/SQLModel's implicit view conversion."
I had a lot of with Code Crafters. It's a paid platform, but they give you a basic walk through of different technologies, with full test suites. For example, you implement some basic Redis. It doesn't spoon feed you what to do, but breaks it down into manageable chunks.
You the mean The War of Art
I went to college late, so I rushed through. Which meant I didn't take the time to really engage with non-CS classes. So I'd like to go back for that. Especially the below masters, which should attract folks with similar feelings.
We're using CodeRabbit at work, and it has been pretty positive overall. Especially once we turned off the poems and sequence diagram noise. Useful enough that I'm looking at paying for it for side projects when I am the sole developer.
I have 2 LG DualUps. Though one is probably more than enough.
From my experience:
Back of House: schedules are pretty much set. The only time things change is working around someone being out.
Front of House is famously a giant pot exceptions. Mix of professional waitstaff and folks who are just picking up some shifts to finance their passion/true focus(art, music, non-profits, teachers). So you'll need to work around some fun priorities.
How do you flag special events? This will require extra people on a Monday that is historically forecast-ed to be slow. Large parties is a specialist skill in a lot of restaurants. Pretty much any server can make a large party work, but normally a few of the staff really shine with that kind of work, and you'll want them staffed.
Respect everyone's availability/time-off: FOH usually has a good mix of full-time and part-time. And a lot of people are willing to pick up an extra shift with some head's up. And that's both part-times going full-time for a week, and full-timers working extra shifts. People have preferences around working/not working doubles and clopens. One of your full-timers requests a few shifts off next week because their band is playing the next town over. The human process is just to ask a few people who are working if they want to pick up those shifts. Often the person taking off will have found someone to cover for them before requesting the time off, so you'll need that input. Often the GM/AGM making the schedule has all of the human parts in their head and just works through it.
This is a large part of the discussions in the first one or two interviews in Interviews with Francis Bacon by David Sylvester. Bacon talks about pushing to the limits of adding more to a work until it's good, and then if taken too far it ruins the work. And only very rarely can he pull it back around to good.
There are 2 different sizes.
Related to wheretodrink.beer, I just launched a rough version of: https://www.nomnominees.com/. A site focused on finding award-winning breweries/restaurants to check out.
Could be a bunch of reddit bots on AWS are now catching back up as AWS recovers and spiking hits to reddit
I'd get it straight from Manning and save a few bucks and take out the middle man: https://www.manning.com/books/build-a-large-language-model-f...
I'm always reading a few books across a categories.
Fiction: Reaper's Gale, book #7 of the fantasy series Malazan Book of the Fallen.
Non-fiction(history): Denmark Vesey’s Garden: Slavery and Memory in the Cradle of the Confederacy
And then I'm dabbling in a few books around the math behind and practical hands-on machine learning/deep learning.
I'm currently working through Mathematics for Machine Learning and Data Science Specialization from Deeplearning.AI. It's been the best into to Linear Algebra I've found. It's worth the $50 a month just for the quizzes, labs, etc. I'm simultaneously working through the book Math and Architectures of Deep Learning, which is helping re-inforce and flesh out the ideas from the course.
[0] https://www.coursera.org/specializations/mathematics-for-mac... [1] https://www.manning.com/books/math-and-architectures-of-deep...
This is where I feel like I'm doing something wrong whenever I try these tools. Everyone says things like "just keep telling it what to fix, it might take a few rounds". But that ends up taking just as long if not longer than doing it myself.
AI doing 30-50% of your work
I use AI for taking info and restructuring it for me. Rewrite a Linear ticket in a proper format, take an info dump and turn it into a spike outcome or ADR doc that I can then refine. I also like it for the rote stuff I haven't memorized the structure of: building OpenSearch queries, writing boto3 snippets, etc. Other than that, my job is the same as it was pre-LLM hype. And from talking to other engineers, it seems that my experience is fairly standard.
I've picked up each color of the new metal body Kuru Toga, and they are an extra level of extravagance on an already perfect pencil: https://www.jetpens.com/Uni-Kuru-Toga-Metal-Mechanical-Penci...
Pigma Microns and Uniball One are my go-to pens. Previously it was Signo DXs as well. I think I prefer the barrel of the DX, but the wire clip and general appearance of the Uniball sold me.
2022 Hyundai Kona N. It's fun while still being practical enough.