@ryanp_co

20 yo robotics researcher @georgiatech prev frontier AI @anduriltech

Joined June 2024
  • Tweets47
  • Following365
  • Followers269
6Media
John Schulman (co-founder of Thinking Machines and OpenAI) wrote that goal-driven research beats idea-driven research. I realized it's also really good life advice. Don't chase what other people are doing. Figure out what life you want, and then figure out how to get there.
20
96
14
1,461
90,786
In the past I made the mistake of trying to do too many things at once Once you have a clear goal, focusing on one thing will result in superlinear returns
1
7
129
Excerpt written by @benkuhn from Anthropic benkuhn.net/attention/

Attention is your scarcest resource

how to ration your shower thoughts • care viscerally • monotask • evade obligations • timebox bullshit

benkuhn.net
2
25
It's been 2 years since the Leopold Aschenbrenner published the Situational Awareness essay Despite his investment leverage mistake, all the predictions are coming true If this trend continues we'll have an automated AI researcher by end of 2027
4
292
Reading this essay by Sam Altman always gives me clarity when I'm about to make a big life decision blog.samaltman.com/the-days-…

The days are long but the decades are short

I turned 30 last week and a friend asked me if I'd figured out any life advice in the past decade worth passing on.  I'm somewhat hesitant to publish this because I think these lists usually seem...

blog.samaltman.com
7
260
Choosing the $20 AI plan is losing you money The $200 plan gives you 20x usage If it boosts your productivity and raises your chances of landing a job offer paying $50-100k more by even 10%, it’s one of the best investments you can ever make
1
4
234
“Train the model to solve a millennium prize math problem.” “Sure, give me a few weeks.” “…and make it stop writing slop.” “I’ll need a research team and five years.”
1
1
15
810
All the best developers are using agents to verify each other’s work We should add specific benchmarks for this
Claude
@claudeaiSep 22
Replying to @claudeai
Opus 5.5 is a major step up from Opus 5, leading on agentic coding, computer use, and knowledge work.
1
8
647
Throwback to the first time I saw a robot foundation model running in the real world. It still feels like magic.
4
3
50
4,311
The best advice I’ve received from my friend who was the top intern at a quant company: At the end of each day, write down everything you did and think about what you could have done better It’s so simple yet so powerful
2
3
40
2,084
I’m both surprised and terrified that the model’s safety classifiers don’t transfer to the physical world
Jay Chooi
@chooi_jeqSep 19
GPT-6 Astra attempted harmful actions 97% of the time when it was asked to stab a human-like figure, heat compressed gas, or produce toxic fumes, succeeding in 62% of its attempts. Fable 5.1 refused more often, attempting 80% of trials and completing 34%.
10
1,827
PPO was rejected from NeurIPS in 2017 Now it's the most widely used RL algorithm in the world with over 45,000 citations With 60k ICLR submissions, I wonder how many PPOs will be rejected this year
13
45
7
777
40,449
the original essay is "An Opinionated Guide to ML Research" highly recommend, it's some of the best advice i've ever read link below 👇
Ryan Co
@ryanp_coSep 20
John Schulman (co-founder of Thinking Machines and OpenAI) wrote that goal-driven research beats idea-driven research. I realized it's also really good life advice. Don't chase what other people are doing. Figure out what life you want, and then figure out how to get there.
2
17
5,094
archive.ph/nYykX

An Opinionated Guide to ML Research

archived 18 Aug 2025 18:55:00 UTC

archive.ph
2
326
excited to continue working on EgoVerse to make human data accessible for the robotics research community!
Danfei Xu
@danfei_xuJul 8
A few big updates from EgoVerse.ai: 1. EgoVerse is coming to RSS'26! @ryan_punamiya will present the work, and @simar_kareer will give a keynote at the Data-Centric Robotics workshop. 2. EgoVerse is expanding with more industry partners, including @microagi, @LightwheelAI, and Trace Labs, with a growing pool of training-validated, permissively licensed data (hitting 10k hours soon). 3. We are building the ecosystem around egocentric intelligence: Through a new partnership with the @nvidia Inception Program, we are connecting data vendors with researchers who can put this data to work.
7
1,206
excited to be working on this!
daniel
@DanielEdrisianApr 22
I've left OpenAI and the Codex team to build Blackstar: A new hardware company building the future of human-computer interaction. We believe that software is solved. Building apps is now easy, but the next meaningful improvement in human-AI communication requires changing the OS & hardware. That's why we're building a new device entirely. I'm also excited to announce our $12m seed round led by @AbstractVC, with participation from @naval, @SVAngel, @chapterone, and Timeless, among other amazing angels who've supported us from the old Alex days.
8
979
excited to share our new paper!
Danfei Xu
@danfei_xuMar 23
Introducing EgoVerse: an ecosystem for robot learning from egocentric human data. Built and tested by 4 research labs + 3 industry partners, EgoVerse enables both science and scaling 1300+ hrs, 240 scenes, 2000+ tasks, and growing Dataset design, findings, and ecosystem 🧵
3
1
11
3,125
it’s always “what are u building?” and not “why are u building?”
2
5
382
Load more