LinkedIn and 3rd parties use essential and non-essential cookies to provide, secure, analyze and improve our Services, and to show you relevant ads (including professional and job ads) on and off LinkedIn. Learn more in our Cookie Policy.
Select Accept to consent or Reject to decline non-essential cookies for this use. You can update your choices at any time in your settings.
South Park Commons is a community of technologists and builders dedicated to helping each other explore, learn, and figure out what's next—or as we call it, go from -1 to 0. We believe great ideas very rarely come about in isolation. The SPC Fund invests in companies that emerge from our members' explorations, ensuring the community's longevity and supporting the next stage of our founders' journeys.
Our South Park Commons BioTech hackathon was full of creative ideas and team names. The Incredible Bowel Detectives won second place for their work identifying personalized biomarkers to predict immune flares. The team of James Newton Brandt,Matt Wallace, and Diana Leung used data from our partner ImYoo to show clear patient-specific gene signatures for flares and a promising direction for future research. Amazing work.
Huge thanks to our partners Amazon Web Services (AWS), Anthropic, Sleuth, OWKIN, Biohub, Phylo, Sid Sijbrandij, GXL, AACR Journals, ImYoo, Cent AI and Mike Snyder's Lab at Stanford.
Very excited to welcome a dear friend to South Park Commons next week.
Sriram Krishnan is joining us for a fireside chat and Q&A.
He helped build Azure, scaled Meta’s mobile ads, and led a16z’s international expansion.
Then he went to the White House and helped write the U.S. AI Action Plan.
Spots to RSVP below.
Dyna Robotics is deploying robots and pushing frontier capabilities. Jason Ma shared his learnings at South Park Commons.
Doing robotics RL research, Jason felt that evaluating his own tasks for papers was disconnected from the real world. So he started Dyna with Lindon and York on one bet. Build the full stack as a product and deploy early, so the real world surfaces the right problems. They started with folding, a dexterous manipulation task that's repeatable across 100K US hotels. It's also forgiving since you cant break a towel. Curriculum learning applied to GTM. Early autonomous deployments in hotels and gyms exposed the next bottleneck. Customers wanted napkins neatly stacked in a bin after folding. So the team built instruction following and reasoning on top of dexterity.
Today the robots run across Din Tai Fung, laundromats, hotels and gyms, so staff who used to fold in the back can serve guests up front. The next bottleneck? The better the robot folds and stacks, the more often someone has to swap the basket. A fast single-task robot still needs a babysitter. So Dyna's moving from tasks to workflows. With Dyna 2.1 and Taku, a dual-arm semi-humanoid on a wheeled base, it can autonomously run an hour-long laundry shift of washing, drying, folding and stacking.
Real-world deployment can be the gradient for what to build next.
Pleasure moderating with Nate Padgett
For years, neural networks looked inscrutable, like tangled knots that needed to be unwound. It turns out this isn’t true: AI models are far more interpretable than we thought. We think this is great news for interpretability.
Goodfire's Nick Cammarata co-authored the paper that introduced polysemanticity and superposition at OpenAI. After months studying neurons in LLMs and vision models, his new essay explains what the early field got wrong – and why there might be a bitter lesson for interpretability.
It’s been an enormous privilege and joy to watch Nick cook this up over the last year. Week after week, experiment after experiment, he unraveled the tangled mess of our thinking about neural nets to reveal the truth. I never knew quite what to expect on our weekly walks. Each conversation brought a torrent of new theories about the nature of AI, some conjectures teetering on the brink of madness, and others that seemed to cut through years of confusion with a single sentence.
Nick is one of the great scientists of our time. His intuition and research taste are out of this world, but what sets him apart is his unmatched ability to capture the essence of what something truly is, without being lost in the details. I would highly recommend you read his entire essay if you’re curious about AI. It's long, but written beautifully.
Nick's research makes me far more optimistic about the future of AI alignment. I hope that it can inspire the same for you.
We're hosting a showcase at South Park Commons NYC to highlight the frontier technologies East Coast founders are building.
Expect live demos across robotics, satellites, foundation models, biotech, AI infrastructure, dev tools, and more.
Every team is hiring, so we're opening up a few spots for people looking for their next thing.
If you're in NYC and want to meet the teams and see what they're building, leave a comment or send me a DM!
At South Park Commons, we are trying to reconcile two things:
1. We have never seen technological progress this rapid or fundamental. Nothing seems impossible.
2. Capital is flowing to consensus teams, in consensus markets, and with consensus terms.
So we’re running a new experiment: $200K, uncapped, to a handful of unapologetically maximalist thinkers who want to try a truly out-there idea.
We want to rewrite the capital stack. If you are building a nuclear company, you probably shouldn’t raise the same seed round as an AI app. One size does not fit all.
Instead, we need to get more people working on ideas that seem impossible.
Use this $200k as no-downside fuel to explore a hard problem, and then we’ll help you accelerate toward the right shape of downstream capital.
We don't know what ideas to ask for. Bring us what you think is most daring and risky. We’re actively looking to rethink all of our priors.
What we do know is that ambition is infectious. Come to SPC, and build alongside brilliant technologists exploring frontiers of their own.
Express interest below.
We hosted Sergey Levine at South Park Commons this week.
He is perhaps the best example of clarity in action that I have seen in a long time.
Thank you for joining us, Sergey.
Last year, in a bizarre turn of events, I got to take Neal Stephenson and Ken Liu—two of our greatest sci-fi authors—on their first Waymo rides.
It was surreal to share a little of the future with the people who imagine it.
The full-circle moment came when Aditya Agarwal and I sat down with Waymo co-CEO Dmitri Dolgov at South Park Commons.
We discuss:
- What does it actually take to remove the human behind the wheel?
- How do you know a self-driving system is actually getting better, and not just getting better at exploiting loopholes?
- Why isn’t a great AI model enough?
"The problem you're working on right now is always the hardest. But also the most interesting."
Dmitri Dolgov has spent his career on a problem most people said would take 5 years. It's been almost 20.
Stanford DARPA team. Founding member of Google's self-driving project. Co-CEO of
Waymo.
When he started, self-driving was still an open question. The jump from making it work to making it safe enough for the real world felt enormous.
At South Park Commons, he talked to me and Ruchi Sanghvi about the 2009 milestones that had nothing to do with building a product. The goal was to learn.
Full conversation at SPC out now.
