Nakshatra Saxena
@nedwize
Building Aida - http://aidabrowser.com • Prev - Product Engineer @Peerlist, @VerakInsurance
Day 13 Update - Aida the browser AI assistant. Aida can now think and execute plans - watch it scrape the current tab for all video titles on @Rahul_J_Mathur's Breakdown channel and save it as a txt file. Note that none of this code is written in advance or just a tool call, it…
Engineers, assemble and try all kinds of prompt injections here please.
Lots DMs asking for free shoes. Let’s do this - follow us and reply with the Gully Labs sneaker screenshot that you want. @grok pick one reply after 48 hours. Happy fun!
Lots of issues with how memory is currently implemented in ChatGPT. It randomly references older chats that aren’t even relevant to the current conversation, and I’ve noticed that it gives less priority to earlier messages within the same conversation compared to those stored in…
Read the Zep paper yesterday. One of the most complex papers I've read around memory, made me feel very stupid. Really liked their implementation of a bitemporal model and also the hierarchical graphs are just top tier thinking.
The actual test of a good marketer now is if they can fly under Phalgun's radar
This job has been taking over my day 😭
Went on a walk alone without earphones on like a true psychopath
You know you're cooking when you get this text.
I like this guy
startup "truth" i disagree with: don't pay high cash comp we now pay 250k-350k base for designers and 300k-1m base for software engineers thread on why everyone is wrong (except me and zuck):
Is this true
melatonin before sleep = DMT demons magnesium before sleep = knock tf out but dehydration l-theanine before sleep = background bliss. floating on clouds. zen garden. forehead kisses. roly poly down the hill.
Excited to share Aryabhatta 1.0, our leading model that scores 90.2% on JEE Mains, outperforming frontier models like o4 mini and Gemini Flash 2.5 Trained by us at @AthenaAgentRL , in collaboration with @physics__wallah, using custom RLVR training on 130K+ curated JEE problems…
Read the original MemGPT paper yesterday, LLM memory being modeled after the operating system is quite fun. A couple of interesting ideas: (1) Virtual Context Management with Queue Managers- Allows the LLM to control its own “virtual” memory and issues warnings at 70% context…
Which memory benchmark do you think is the most representative of real world usage?
More people should create things to proliferate an aesthetic into our future, not just to solve problems. This is the quality that every artist and engineer I respect shares most universally. Without this, you are doomed to churning out slop.
More people should create things to proliferate an aesthetic into our future, not just to solve problems. This is the quality that every artist and engineer I respect shares most universally. Without this, you are doomed to churning out slop.
great people just take good people who are hungry to be great. they throw them into situations where they can be great. and that's when good becomes great.
