polyphilz
Re:Training a 4B model to produce 81% faster query plans than Postgres
on Wednesday September 16, 2026 @03:40PM (#49731869)
`pg_hint_plan` has a debug log so you can verify Postgres actually used the hint or not! Used this during evaluations
Conviction
Re:Show HN: Neolabs.fyi – 100 new AI labs by research area, valuation,...
on Monday August 10, 2026 @12:34PM (#49246032)
Thanks! Funding that's public is used directly, else it's marked as undisclosed and the size of the hex for that lab is inferred based on common multiples. Right now it's static with the goal of moving to something that can propose changes as news is released
Show HN: Neolabs.fyi – 100 new AI labs by research area, valuation, and more
Ask HN: What's SoTA in Agent Memory?
I built a Telegram agent for my partner and I where we can message it photos or descriptions of food we had, and it logs our macros. I wanted to build a "food memory" feature, where I could say something like "had my usual smoothie" and it remembers the macro breakdown of it.Curious on what the best approaches are nowadays for this type of agent memory? Papers, articles, tweets or github links...
Show HN: Glance – Local Git diff review TUI, ported from VS Code
Show HN: Ccnotifs – macOS Claude Code Terminal Notifications (Tmux Friendly)
https://github.com/polyphilz/ccnotifs
A dead-simple macOS notification system for Claude Code.Something I've been meaning to build for a long time. Uses Claude Code hooks (Notification and Stop events) to shoot you native notifications when your input is needed, or when Claude Code has finished and is awaiting further instructions.Supports tmux out of the box: clicking a notification focuses your terminal and chains tmux commands...
The AI bubble argument misunderstands both bubbles and AI
Re:Ask HN: The government of my country blocked VPN access. What shoul...
on Friday August 29, 2025 @04:05PM (#45068784)
Experimented a bit with ChatGPT and it seems to freaks out at the "bypass censorship" language in particular. I re-framed the request more around helping me understand networking better, and it complied immediately
Travis Kalanick wants to push state of the art in Backgammon
Ask HN: How do I become a better PM as an engineer?
Context - I've been a software engineer for ~5 years now and just joined a seed-stage startup as a founding engineer. Of course, founding engineer != PM, but we don't have a PM and I learned (the hard way) recently that I'm not great at putting the product hat on. How do I get better at understanding the needs of my users, writing PRDs and all the other work that goes into being a good PM at an...
Family of beavers save Czech government $1.2M by finishing dam
Glue Work Considered Harmful
Out of the Fire(store): Traba's Journey to Postgres
Re:The Leverage of LLMs for Individuals
on Thursday May 11, 2023 @04:44PM (#35908106)
> Staying in any company right now is a negative return; you are wasting personal leverage.
I left my job at big tech a few months ago and started using ChatGPT (and now GPT 4) shortly after for personal projects.
100% stand by the author's statement. The ability to iterate and execute on my ideas in a way I've never been able to do before—combined with not having to worry about whether I'm infringing on my previous employer's intellectual property rights clause—feels so incredibly freeing.