Supreme Leader Wiggum

If I were to stream… and it was mostly just me doing my day to day work, while talking to chat, would anyone watch? Setup isn’t going to be anything stellar, no real agenda - I’d pretty much just go do stuff across 5 or 6 repos in parallel & talk through what I’m thinking

@davis7 1 agent per token

- 1 agent per file is really nice for refactor stuff - 1 agent for each review note codex left to verify and fix - monitoring CI - research subagent works better than expected - parallizing work within one big job (do ___ to each of these clips)

@wordmandotdev @intellijidea 🤣

@ScriptedAlchemy @intellijidea lol it needed an hour to generate the answer 🤣

@wordmandotdev @intellijidea Says decade plus, does it in 3 hours.

@alchemysupplies itll be better.

@ScriptedAlchemy Think it will be on par w Fable? I'm a Codex stan but Fable was on a completely whole new level

Literally have things parked, waiting for GPT 5.6 to be released. Not even worth burning the tokens to try and do it now.

Weird new type of spam on this app, dozens of accounts impersonating me that retweet every post I make

get rid of your humans and lets your infra team talk to the humans agents instead :P

@jiahan_c bahahaha aint that the truth

@ScriptedAlchemy The best part is that Codex gives enough context to describe the whole issue. In my experience, getting humans to do that is the hard part 😄

Very funny situation. Im debugging build issues for a user. We basically have our instances of codex talking to each other, the humans are just the ones in slack to relay the messages between our codex's.

@CaddyComa i think complexity has ticked up in recent years but hope to simplify it further in near future

@ScriptedAlchemy What's really admirable is how good this stuff works in production. It just works™ Great job simplifying this en masse, it takes a lot of effort to get to such a distilled abstraction. KUDOS

It’s actually insane 10 million weekly’s. More downloads than angular

@wesbos Xiaomi means “small rice” in Chinese

xiaomi - the Chinese company that makes phones, rice cookers and electric vehicles - has forked OpenCode

A memorable moment! Module Federation Runtime has surpassed 10 million weekly downloads 🎉 Thank you for using our code.❤️

@yabatrades @tlakomy

@ScriptedAlchemy @tlakomy Literally the only thing my manager knows AI related are the words cursor, claude and open ai. Nothing past that. Saw reports of a whole other dev team under her using haiku as of last week (since late last year) It is an absolute shit show in large enterprise

I still can’t believe we used to sit in conference rooms and estimate (lol) tickets in “story points” instead of … shipping stuff. We can never go back.

@thsottiaux @SourabhGurwani

if name == "sourabh.gurwani": return False And every time a user registers, you start a Codex automation job to add a new if clause (very quickly). If you want to optimize you can use GPT-5.3-Spark, which is faster. If you want to optimize further, ask Codex how to implement an index.

Interviewer: You enter a Gmail username and the UI immediately shows "Username already taken" There are millions of users worldwide. How does it respond so quickly?

The Rust React Compiler integration in SWC has been merged 🎉 Coming soon in Rspack 2.1

The backtest uses online updates. So I label the data. Train model. Then use point in time replay, so backtesting it does online training since that’s just part of the system. If I just used the same checkpoint and Ran the same backtest twice, the second time it would be improved due to online parts.

@ScriptedAlchemy That makes sense. Last thing I’m curious about: do you train from simulated labels only after the full 2-6 week path is known, or do you also do online updates from interim mark-to-market premium path outcomes before the trade would have exited?

Fable stuck around just long enough to debug my 4 billion rows of training data, train a new financial model that was twice as profitable with half the drawdown, and during stress periods, maintain profitability. The model is now finally passing the 30% a month yield.

Use custom universe currently since I need training data on something. Opra data backfill is massive. Terabytes. So I use universe, then I have separate project that was experimental where I just use recommendation engine and bluesky firehose to rank accounts by accuracy and lead time. Then use that to build reputation database and discover instruments through social data for potential use. If don’t have to train on it all, then I’d just open socket connection that watches top 500 or top 1000 stocks by market cap. Let online training work it out in sparse heads/prime it with historical

@ScriptedAlchemy How are you finding the Stocks to watch? Do you scan a custom universe daily? Also any chance you share it?

@HashemKhalifa Model figures out weighting on its own. I create real simulation using OPRA quotes and trades, with slippage and latency. Little approximation since I have the order book. Lots of data labeling materialization. Watch for over fitting. Never test on data you trained on.

@ScriptedAlchemy One follow-up if you’re willing: when you optimized for 30% monthly return, did you train that as a direct portfolio simulation objective, or use weighted per-example labels that approximate monthly capital-cycle return?