Senior Software Engineer @metropolisio | Building the Recognition Economy
Joined August 2023
- Tweets 688
- Following 315
- Followers 57
- Likes 1,854
60 Photos and videos
Jun 11
You're in the danger zone with LLMs when you're actioning on their output but don't understand the thing they're working on.
2
Jun 10
I know we're in 2026 here, but I want to say @springboot JPA, hibernate, whatever you want to call it, is horrible. Why must my DB model affect every layer of my application. Why would I want to tightly couple my service layer code to my db table? It's horrendous. Beyond me why it's the default.
1
20
Jun 10
Timeline is hot takes from folks about fable that I don’t even think they themselves believe
9
Jun 10
Spent 20 dollars using Fable to create a Jira ticket that nobody is ever going to do
17
Jun 8
IDK if I'm getting worst at reading code or LLMs just suck at writing it
1
17
Jun 8
"Claude let's think through some architectural decisions. I think we could do x or y."
"Do x."
"But x has this problem."
"You're right do y."
"But y has this problem."
"Ah true do x then."
Thanks for nothin Claude.
2
21
Jun 8
As engineers our job is to find and reduce complexity. LLMs are not very good at that job.
2
8
Jun 7
Using anything besides Claude Code, makes you realize how bad Claude Code is at managing context bloat
1
20
Jun 7
Extending hermes to get plots out of the groupchat with the boys. It'll find deals, gauge interest, align schedules, and plan itinerarys
28
Jun 7
Seeing if I can get my hermes bot to spin up a gpu pod and generate dubstep songs for the homies in Discord. Ty @runpod
14
Jun 7
The only thing stopping me from believing that @AnthropicAI is completely delusional is the @bunjavascript rust rewrite. Even then most agentic workflows don’t have a phenomenal language agnostic test suite.
34
Jun 7
Sometimes I game with the homies. Instead of just sitting and coding.
17
Jun 7
What models are people using with @opencode? Seeing how long I can go without a ChatGPT subscription
1
33
Jun 7
Lately been using deepseek v4 flash. Unfortunately most of my work relates to tiananmen square.
27
Jun 7
So I guess open router only auto routes on first prompt. After that you’re pinned to whatever the model choice was.
Jun 6
No! when you get cached by a model or provider, you get pinned to it until the cache expires. You can also set explicit caching. See openrouter.ai/docs/guides/ro…
52