I also have scripts to fetch specific database assets and forward them to slack channels so I can easily share them with a group rather than manually running a query and generating them.

I had a theory about improving a product. I asked it to build an offline simulation setup to try various implementations. The results were a bit fishy but i decided to give it a try and A/B testing is showing similar results.

And now im vibecoding a locally hosted dashboard. This one is less useful for anything specific, and more of a minor quality of life improvement, but its fun to just vibe code and see changes happen occasionally. Its not a critical thing.

I attempted to run the project entirely asynchronous, where we had a slack channel, ICs had their section of goals and milestones, and I was there for them to consult with, provide feedback, unblock obstacles, and proactively come up with interfaces across the objectives. I thought this would be a nice high trust method of doing things that gave people ownership over their respective parts.

What happened was one person made an AI copy of the doc I had and began vending that out to everyone else, progress was quite slow and really complex in original proposed PRs (unsure if thats AI or author doing that), people did not really follow through with their implementations and it all ended up taking longer than I expected for no good reason. In the end, I lost trust in these ICs as I now feel the need to chase them and have low desire to work with them.

For the next XFN project, I will be driving a brief weekly meeting. Unfortunately the pressure seems to be important. I think there are things I could've done better communication wise at the beginning and throughout as well, but overall I felt disappointed that I had to check in to see progress.

Even if every employee fired saved $500k a year, that'd be roughly $4B in a year. Not a small amount but relative to their income, not huge either

We got broad and wide access to AI tools maybe a month ago now. AI tools meaning claude code, codex, cursor and a set of other random AI tools.

I use them very often. They've taken a lot of the fun and relaxing parts of my job away and have overall increased my stress. I am on the product side of the business and it feels necessary for me to have 10 new ideas and now the ones with the most ideas will be rewarded, which I am not as good at. Ive tried having the agents identify opportunities for infra improvements and had no good luck there. I haven't tried it for product suggestions but I think it would be poor at that too.

I get sent huge PRs and huge docs now that I wasnt sent before with pressure to accept them as is.

I write code much faster but commit it at the same pace due to reviews taking so long. I still generate single task PRs to keep them reviewable and do my own thorough review before hand. I always have an idea in ny head about how it should work before getting started, and I push the agent to use my approach. The AI tools are good at catching small bugs, like mutating things across threads. I like to use it to generate plans for implementation (that only I and the bots read, I still handwrite docs that are broadly shared and referenced).

Overall, AI has me nervous. Primarily because it does the parts that I like very well and has me spending a higher portion of my job on the things I dont like or find more tiresome.

Maybe OpenAI does things different, but as soon as an OKR around ad performance gets committed to, the experience will degrade. Sure they're not selling data, however they'll almost certainly have a direct response communication where advertisers tell Open AI what and when youve interacted with their products. Ads will be placed and displayed in increasingly more aggressive positions, although it'll start out non intrusive.

Im curious how their targeting will work and how much control they'll give advertisers to start. Will they allow businesses of all sizes? Will they allow advertisers to control how their ads work? I bet Amazon is foaming at the mouth to get their products fed into chat gpt results.

Its awesome that youre feeling better. Its possible, but hard to believe, that its due to nothing but diet changes and if it is, then its hard to imagine that such an extremely specific diet is needed to get the same results.

I use multi armed bandits a lot at work, and I wonder what other algorithms I should look into. Its hard to read the name of an algorithm or category and get a sense of whether or not its applicable. It seems like the general process of "which path out of N options is the best?" can be reframed in many ways depending on contextual details such as how large is N, what's the feedback latency, what are the constraints around evaluation and exploration, etc

* we have a single optimization goal (e.g the metric to minimize or maximize). The hard part isnt defining a optimization goal. The hard part is identifying what stakeholders actually want the tradeoff between different metrics. If you have goal A and goal B, where B is more aggressive than A, then getting agreement on which is better is hard. This is a people problem.

* MAB seems to be a good proof of concept that something that can be optimized but isnt an "end game" optimization path.

* MAB for A/B testing is going to mess up your AB data and make everything more difficult. You should have a treatment that uses the MAB algorithm and a treatment that doesnt.

All of the above is for non contextual MAB. I am currently learning about different MAB algorithms although each of them are pretty similar. The ones ive read about are all effectively linear/logistic regression and tbe differences come from exploration mechanism and how uncertainty is represented. Epsilon greedy has no uncertainty, exploration just happens a fixed percentage of time. UCB is optimistic about uncertainty, and the amount of optimism controls exploration. Thompson sampling uses statistical (beta) distributions about uncertainty, exploration happens less as confidence about a particular set of options increases.

Overall its a fun area to work in that is quite different from your typical CRUD development which is a nice change of pace.

> you’d have to either memorize the entire answer before speaking or come up with a simple pattern you could do while reciting that takes significantly less brainpower

This part i dont understand. Why would coming up with an algorithm (e.g. a simple pattern) and reciting it be impossible? The paper doesnt mention the models coming up with the algorithm at all AFAIK. If the model was able to come up with the pattern required to solve the puzzles and then also execute (e.g. recite) the pattern, then that'd show understanding. However the models didn't. So if the model can answer the same question for small inputs, but not for big inputs, then doesnt that imply the model is not finding a pattern for solving the answer but is more likely pulling from memory? Like, if the model could tell you fibbonaci numbers when n=5 but not when n=10, that'd imply the numbers are memorized and the pattern for generation of numbers is not understood.

This is what I've found it most helpful for. Typically I want an example specific to my scenario and use an LLM to generate the scenario that I ask questions about. It helps me go from understanding a process at a high level, to learning more about what components are involved at a lower level which let's me then go do more research on those components elsewhere.

Do companies like this have a need for SWEs? Are there opportunities for a backend SWE without any background in hardware or biology?