GPU Poor? **Friends** is all you need.
i built a distributed training module from scratch that performs model training over a cluster of M-series macs connected over the same network.
this was inspired by
@anyscalecompute
's Ray,
@hnasr
's work,
@ShopifyEng
's merlin article and
Introducing BeyondPDF, a smart, free, fully-local (wifi-less) PDF search app for MacOS. "Free" has always connoted to "It's free but we will steal your data"... not anymore.
Whats wrong with MacOS's Preview? It's great... but I NEED smart/semantic search. Searching for words is
gpu poor? I wrote a simple distributed training module to train a vision model on three M1 macbooks. it's open sourced!
i wanted to build a distributed training tool for M-series macs but I faced two hurdles: network latency is extremely high and my friends weren't happy with
AI chatbot startups have raised $100M+ right now but have it all wrong.
Your chatbots point users to a 100-page zendesk doc on "how to ***" but why not just complete the task?
I built ActBot: set up your own agentic chatbot on any website!
The demo below shows ActBot
Introducing Agenata: a free, open-source web annotation tool. The demo below shows a live annotation of adding an item to cart on uber eats.
Currently, SOTA web agents underperform because:
1) Fewer web trajectory datasets lead to poor reasoning and planning over different
I built an AI agent that can perform end-to-end tests on your webapp with natural language test definitions.
traditional heuristic evaluations cost $10,000+ dollars, an llm-agent-based heuristic eval can cost < $1,000, even for large projects!
shoutout to
@ivan_yevenko
from
Self-building, on-the-fly APIs with agents. Demo ⬇️
The world needs devs to spend more time architecting developer-friendly APIs not code monkeying the same boilerplate.
I time-boxed myself to build a simple self-building, and soon self-healing, API that can build and fix
I built a patent search engine to look through billions of patents (my demo only does millions for now) through semantic search.
The infra for this is very hard and required some work on the embeddings / vectordb front.
To scale up your embedding search, you need:
- small
@sincethestudy
@natfriedman
yo brian, thank you for helping me build the thermoclinic neucliophillic ai agent reactor that, since the study, has helped cure cancer (currently in stealth mode)
Hey
@metaphorsystems
(
@wangzjeff
@SarahChieng
)! I use google and metaphor frequently, and recently, I have been thinking about how Query Biased Snippets on Metaphor's search results could really enhance user experience.
Here's a quick video/demo of metaphor with QBS.
Things to
Introducing BeyondPDF, a smart, free, fully-local (wifi-less) PDF search app for MacOS. "Free" has always connoted to "It's free but we will steal your data"... not anymore.
Whats wrong with MacOS's Preview? It's great... but I NEED smart/semantic search. Searching for words is
shoutout to
@globe_engineer
, i took their DOM optimizations
shoutout to
@tryramp
and
@karimatiyeh
for beating me to build this
this is super crude but works well -
i know a waterloo first year still looking for a summer internship. good full-stack, AR and embedded experience but most importantly, done to learn. anyone interested in a chat?
Just got word that my fall tesla internship on the battery manufacturing eng team was cancelled due to layoffs. I'm a mechatronics student at uwaterloo, if anyone is looking for a mechanical/robotics position, let's chat! down to grind on the most ambitious ideas :)
im SO bullish on the mlx. when apple builds, they build seamless integration within their ecosystem. Federated Learning is great across mac mps backends, but.... I have some thoughts on this, but I will let you extrapolate :) maybe I'll even build and OS it
If a service is free, you are the product. gpt was never meant to have just a consumer biz model, it does not scale as well... now enterprises (like apple) will fill oai's pockets for us to use it... parallels with how Google provides search for free. giving it for free leads to
my intention is to get this to be readable, DX-oriented and open source it so people can learn that federated learning is actually not hard (on a dev scale)
Introducing Ramp Tour Guide: an AI Agent that can show you how to do anything on Ramp!
Today, we'd like to share a sneak peek of Ramp's near future. As Ramp grows in functionality, we want to make all of it easily accessible to all of our customers. To do that, we're demoing a
@cHHillee
Also its an inference chip... this is great competition for groq not the H100, im hoping a distributed inference provider steps in and provides great competition!
I made a big list of books based on my last tweet's replies which ill pick from next time. I got the one below. ever since i became more ambitious (2nd year waterloo eng onwards), I have felt a deep difference in friendships I had in high school vs rn. I realized I became
im going book shopping, drop your favorite book recs!
idc if its about business, fiction, science, biographies, etc. I want to expose myself to something new. will post an update of what I buy :)
apple intelligence will be apple's biggest feature. we talk about why agents should not use user interfaces and instead get work done with api calls to services... but most services don't actually offer public apis. through app intents, apple basically gets their top apps
@VictorTaelin
twitter dms and replies for hiring is noisy, two ways iv seen ppl make great hires:
- approach a highly technical vc (they exist) and get them to make intros within circles (they know ppl at nvidia, meta AI, etc who are bored out of their minds wanting to work on next gen
im going book shopping, drop your favorite book recs!
idc if its about business, fiction, science, biographies, etc. I want to expose myself to something new. will post an update of what I buy :)
decided last night that grpc is not the right tool to use on my simple-federated-learning tutorial repository (and is a pain to build on). its going to limit the repo's power in penetrating the depths of your intuition. switching to basic REST.
Introducing Agenata: a free, open-source web annotation tool. The demo below shows a live annotation of adding an item to cart on uber eats.
Currently, SOTA web agents underperform because:
1) Fewer web trajectory datasets lead to poor reasoning and planning over different
the openai's and deepmind's of the world are in a special position. most companies have to improve ux, add features, etc to win over competition. oai, dm, etc have to perform and invent literal magic to win over each other...
won (with my team) the
@UWVelocity
agetech challenge by building a working prototype that can help detect early signs of dementia and costs < $10 to build
currently such machines cost $500,000+ and work by measuring plaque deposits in retina
looking to build this out further!
yesterday I completed a dist sys project in 12 hours that I had failed to accomplish in two previous attempts each of which took more than a day. it did take me a year to build the intuition. if people like it upon release I will build this further for fun maybe even drop a tut
imo openai is economically constrained from releasing gpt-5. compute is a big bottleneck and they havent figured out how to effectively reduce gpt4's size to support offering that efficiently. but satya could step in and just eat the costs so openai can dominate. EOD, jensen
the tech community needs better fake praises like
- you have the "creativity of an ai founder"
- wow you have the "MMLU of gpt-3.5-turbo"
- dude you as as agentic as SOTA ai agents!!!
"Introducing MAMO. <=>"
"MAMO is a multimodal interface that lets you talk to your computer without needing to make a sound. <=>"
"Powered by
@symphoniclabs
multimodal Silent Speech Foundation model. <=>"
Signup for MAMO beta here:
cc
@_nightsweekends
I sometimes imagine myself on my deathbed and reflect on "biggest regret?", etc type questions. starting to realize any answers I think of now or would say then, are going to be very biased. Even answers from people currently on their deathbeds will be biased by nature.