bucket2015

As far as I can tell, it's just a neat Sci-Fi short story from 2016.

I find that if I leave an instruction in AGENTS.md to "do not do X", there's a good chance the agent will forget it.

But if I add a separate post-implementation pass to "find and fix X" by the agent, it'll usually find and fix the issues.

So I've started doing it for everything from naming conventions to duplicate code to other problems. It does cost more tokens, but now I get less frustrated at having to fix basic issues in the PRs.

That's a fair point. I guess step 0 is that you have to care about code/product quality and prioritize it.

It's hard to suggest something without knowing what you already do or what you like. But here are a couple thoughts for you:

* I'd guess that for the next ~5 years at least, the world will still need software engineers. AI may be good at coding, but it's still not great at software engineering or managing software delivery.

* Most non-technical people would still likely prefer that someone experienced packages a system for them. They don't have time, energy, or knowledge to babysit an agent. They may not even know what or how to ask.

* It's easier to write software than ever. This competes with coders (possibly putting jobs at risk), but complements those who create useful things. So focus on being the person who solves problems and not one who just builds code.

It's interesting how there's an increasing split of software engineers into those who're very happy with AI and leaning more into it, and those who're increasingly skeptical or dejected.

I don't know how this will play out, but I haven't seen anything like this before.

That's actually the thought I ended up having: could there be asymptomatic rabies that never go noticed or reported?

Anecdotally, I know several companies that have given the directive of "Use more AI instead of hiring more people". I wonder whether this is part of the story here too?

For anyone confused what "RAG" is: it stands for "Retrieval Augmented Generation", which is where the LLM model is paired with some sort of "database" (often a vector database) where LLM looks up additional data when performing the task.

I was looking at this recently. Although duct tape doesn't always win vs cryotherapy, it's definitely a valid treatment:

* Silver duct tape occlusion in treatment of plantar warts in adults: Is it effective? (https://onlinelibrary.wiley.com/doi/abs/10.1111/dth.13342) found that cryotherapy outperformed duct tape over a 8-week treatment period.

* Duct Tape Occlusion Therapy in the Treatment of Plantar Warts (https://conservancy.umn.edu/items/4e958e09-839b-4496-972f-cf...) is a case study of treating one one patient with large plantar wart.

* Successful treatment of refractory plantar wart with duct tape (https://www.jaad.org/article/S0190-9622(21)01507-3/abstract) is another similar case of treating a refractory (treatment-resistant) wart with a duct tape.

Yeah, this was pretty fun!

Here's rap about a duck trying to get chicken nuggets at McDonalds: https://app.suno.ai/song/27476a1c-ab35-4dbd-bf71-a9fa0aaa4de...

I don't see a way to make anything longer than 20-30 seconds though.. or is that limited by the lyrics?

Edit: with help of ChatGPT I got some more lyrics, but looks like this is as far as Suno.ai is willing to take it: https://app.suno.ai/song/1694f2a9-1375-4dc9-b523-33528aa308a...

OP here - Thanks! I'll put back the optics courses I deleted then

These are awesome recommendations - thanks!

That's a good point. The project-first approach worked very well for me when learning new software frameworks/libs/etc.

My main concern with EE is that once I'll get to the brain-computer interfaces, I'll be in a situation where there aren't many off-the-shelf components/solutions available, and at the same time I'll likely need to know how I can push physics closer to the edge. I suspect I may need a better theoretical foundation to do that.

That said, I definitely like the idea of focusing a lot on hands-on projects.

About 1/2 of the textbooks came from the old course outlines (e.g. https://ece.uwaterloo.ca/~ece207/) and class shopping lists on UWaterloo book store. About 1/4 were guesses, and the other 1/4 I don't know where to even start looking.

The higher the course the lower the confidence of having the right book.

OP here. Nice, thanks! I didn't put math down because my undergrad was in applied math and I got to use it a lot early in my career. But sounds like I'll definitely need a refresher.

Yeah, my first guess is the movement of that faint inner line triggers the Parasol cells or some other movement detection cells in our retina (https://en.wikipedia.org/wiki/Parasol_cell).

Possible underlying circuit: https://www.cell.com/neuron/pdf/S0896-627318%2930105-3.pdf

Vena Solutions | Various engineering openings | Full-time, Onsite | Toronto, Canada | https://careers.venasolutions.com/job-board/#

Vena is a Toronto-based rapidly growing startup with 300+ employees. We build Financial Planning and Analysis (FP&A) software that's based on an Excel plugin.

Looking for developers of all seniority levels to help build new features as well as solve back-end and front-end challenges as we scale the solution to more customers and more data.

I don't think it's feasible to have a satellite orbit at 100 km and not fall out of the sky in a few days due to atmospheric drag.

Even ISS at 300-400 km orbit has to be regularly reboosted because it's continuously losing altitude.

What's "BS" and "MS" here?

> 11/18: Started going through the French Duolingo tree

> 3/19: Reached the end of the Duolingo tree (level 1 in all skills, but most were level 3+)

This is pretty impressive. I've started French Duolingo around the same time and have been doing 30 min per day every day; 14 months later I'm about 1/2-way down.

This must have taken ~3h/day of Duolingo.

I've been using an Excel file (https://i-kh.net/2019/10/20/how-i-keep-the-side-project-movi...) for week-to-week goals and tracking how I do.

I find it's a bit more structured, and easier to review afterwards. Though I still use text file for organizing a day's work.

Yeah, from personal experience (and from watching others), a lot of the habits like the ones the author cites last a few months and then you end up dropping them for whatever reason.

That said, I did manage to stick to an explicit system for a bit over a year now which produced some imperfect but adequate results: https://blog.maesure.com/2020/02/01/anti-procrastination-sys... (disclaimer: my product blog)

I have a bunch of side-projects that I've been working on for a while.

To be honest, I prefer sticking with major cloud vendors (currently on GCP) because:

1. They already have reasonably straightforward ways to publish apps, e.g. I use GCP App Engine + managed Postgres DB + storage buckets.

2. If my project grows or pivots, I like having an option to use other cloud provider services like queues, BigQuery, VPC networking, etc.

I recently quit coffee and was also surprised to get more energy (after the headache was gone), especially in the late afternoon and evening.

That's really unfortunate since I really enjoy the taste of coffee.

That's pretty cool! This is pretty simple and minimalistic.

I've been working on my own different take on time tracking too (https://maesure.com) which asks you periodically what you're doing and builds a picture of the day from that.

Pretty cool - but the music is rather distracting.

My company (a large-ish pension plan, ~10 dev teams, most applications are internal) was on AWS Beanstalk and switched to Kubernetes (EKS). Once we moved in and cleared the initial hurdles, the overall impression is that K8s is very pleasant to work with: solid, stable, fast, no bad surprises. Everything just works. We probably spend 0.1 FTE taking care of Kubernetes now. Definitely was worth the cost.

All AWS tech I've tried before (ECS, Beanstalk, plain EC2, Cloud Formation) is slower, has random quirks, and needs an extra layer of duct tape.

Fullstory basically lets me sit behind the shoulder of every user and e.g. watch them use the app in a way I didn't think of, or stumble over the features that I thought were obvious but turns out are very counter-intuitive.

It's totally a game changer for UX design.

Heap, Google Analytics, etc are actually not as useful; I find the app's DB and server logs provide more relevant data. So I don't think I'll add them back.

Fixed.

Fixed