spicypunk
@bethgenewang
playing an infinite repeated game use ML to teach me learning @columbia see u @_theresidency
new Build-in-Public trend I saw step 1: build nothing step 2: call yourself a $$$ guru step 3: post about it step 4: get 10k likes
Earlier this week, I talked to 100 strangers in NYC in a day on camera — with @lilianbuilds and my brother. Here’s what I learned: 🧵 1. Rejection isn’t about you Most people said no because they were rushed or didn’t want to be on camera. It had nothing to do with me. ✨ Just…
got stuck chasing the perfect app revamp last week almost ended up ditching the project –– perfection kills execution today I just focused on learning one piece of tech —ended up doing the most (time blocked for sunset walk)
your ambition and hardworking-ness is the average of the 3 people you spend the most time with – use availability bias wisely
Instead of a mindless "what are you building?" I'm gonna ask "what interests you recently?" 😉 I genuinely thinks people are more than the things they work on atm, and the interest question opens more space for conversation and future collaboration
“What are you building?” Ahh text
Last night I simply could not sleep can't believe I'm gonna to see the blog author @sama for a Q&A in 30 min you can bend the world to your will a surprising percentage of the time
It's crazy how adding a more direct CTA for my app 300% the downloads I got. I used to think the CTA in the video makes the ads too conspicuous. But if your app solves a real problem, people need a direct nudge 👉 Even tho other video got more views and likes, their…
If you wanna join me at the coolest Gala @_TheResidency in SF this Sat, DM me for invitation code
have been daily posting on tiktok for ~a week So far, 6.4K views on 7 videos and 337 likes -> 10 downloads + 0 users on subscription 😅 should I keep finding the winning tt format or tweak (remove) the hard paywall? suspecting my data is too small to tell
Caught myself spiraling after hours stuck on model output📉 Inner critic was loud: “Is this even going to work?” NoCap catches my inner doubt ☮️ Catch-> Reframe -> Reset
Do people use Langchain to make the openAI response more reliable? I've trying to produce more consistent response from OpenAI API (e.g. consistent depth and thoughtfulness of the answer). Curious what are your approaches. 🙏