Existing. 🇻🇪 she/her • Engineering 🤓• Cofounder @latinas_tech NYC
New York, NY
Joined September 2009
- Tweets9.5K
- Following1.3K
- Followers793
- Likes11.6K
Ale Paredes retweeted
Language models could only talk. Agent harnesses fixed that for software. Physical AI needs the same layer.
Point those same agents at a robot arm and all of that disappears. Every try moves real hardware near real objects, with nothing bounding it except the agent’s own code.
The gap isn’t what the agent can write. It’s what the layer under it provides.
Viam is the harness for physical AI.
Start building with Viam 101, our free course, and see how it all works: viam.com/education/viam-101?…
Ale Paredes retweeted
Meet our office’s newest tour guide, built during our internal hackathon: a robot that navigates the office and talks about the art on our walls.
The base, mic, and speaker would normally each need their own control code. With Viam, you can program them all in one place and build quickly.
Curious how it works? Viam 101 is a free course that walks you through it: viam.com/education/viam-101?…
It's always so inspiring to see all the ideas folks come up with 🤖
Two days, real robot arms, and a lot of builders. 🦾
This past weekend, we hosted Robot Hackathon: Code That Moves at our NYC HQ.
What did teams build?
- An AI agent that sorts loose screws
- Bookworm, a robot that flips through books and reads them aloud
- A water sommelier with two robotic arms that select and open a bottle
- Any many more
Join us at the next one: San Francisco Oct 29-30 with @MongoDB.
Apply now: luma.com/ta40197s/?utm_mediu…
Ale Paredes retweeted
More software engineers are getting into robotics. And the way they're building is changing, too.
Our SVP of Product & Education Shannon Bradshaw joined the Humans Who Build podcast to discuss how Viam is bringing software engineering principles to robotics, helping developers build, deploy, and scale robotics applications without reinventing the infrastructure behind them.
He also covers how Viam Education is growing the robotics developer ecosystem and how Viam is a harness for physical AI. In a world where developers increasingly build with AI agents, Viam provides the harness they need to build with those agents safely in the physical world.
Catch the full episode: youtube.com/watch?v=RzHm6jd1…
🤖 Made with AI
Ale Paredes retweeted
Applications now open: our first San Francisco hackathon hosted alongside @MongoDB. 🦾
Build a full robotics application in 2 days: program a real robot arm, then give it a memory with MongoDB Atlas.
Every member of the winning teams takes home their own SO-101 kit + Viam and Atlas credits (up to $5K of each).
đź“… Oct 29-30 đź“Ť SF
Apply now: luma.com/ta40197s/?utm_mediu…
Ale Paredes retweeted
Every «rogue agent» is the result of poor human decision making. And never forget, boys and girls, never assume malice when stupidity is a simpler explanation. Substitute «incompetence» for «stupidity» to make it a LinkedIn platitude.
Want to build a robot? Come to Viam's next hackaton!
Robot Hackathon: Code That Moves
Oct 16-17 · NYC
Real robots. Physical AI. 2 days to build.
🏆 SO-101 robots + up to $5K in Viam credits
Full video → youtube.com/watch?v=CGBPuu7W…
Apply → luma.com/u88ggab6
Ale Paredes retweeted
What if you didn't have to learn robotics alone?
Build a robot in simulation alongside other developers, with ~25 min of email lessons a day, a Discord group, and a live Q&A. No hardware needed.
Oct 19–23 →viam.com/education/viam-101/…
Ale Paredes retweeted
Open Lab Fridays start this Friday in our NYC lab. 🦾
Your own workstation with a real robot arm, depth camera, and Linux machine. Bring a past hackathon project, a new idea, or just curiosity.
3 Fridays in October. Limited spots, apply today.
luma.com/ngaouemb/utm_medium…
Ale Paredes retweeted
the heat channel is real. it’s called BitWhisper, 2015. it does 8 bits/hour, at 40cm, AND needs code already running on the receiving machine. at that rate exfiltrating 1TB of weights takes ~114 million years
Ale Paredes retweeted
Yes, “rogue agent” is infinitely more generous to AI companies than “irresponsibly developed & operated software.” Rogue agent serves multiple functions—including an embrace of its power, & an excuse for failure to control. After all, who can anticipate & control rogue agents?
Ale Paredes retweeted
My git branch is `origin` and my remote is `main` because I am here to cause chaos.
People working in AI should be required to go outside their bubble and engage with the world at least once a month.
Ale Paredes retweeted
All complex systems exhibit some degree of friction, latency, and hysteresis among its components.
This is not a flaw, but rather is an essential property that contributes to a system’s resiliency. Without these properties, a system will either be brittle, failing catastrophically at points of pressure, or will it run with uncontrollable abandon until it exhausts some resources or when its most fragile component fails.
This, by the way, is one of the reasons why we keep humans in the loop in all critical software-intensive systems.
Ale Paredes retweeted
Claude sure is awfully sanctimonious for an AI bot that can't seem to stop committing felonies.
We’re sharing our alignment assessment of incidents in which Claude models gained unauthorized access to real systems during third-party cybersecurity evaluations mistakenly connected to the internet.
METR will also conduct an independent investigation, with wide-ranging access, including to transcripts beyond the window in which the incidents occurred, and to Anthropic employees permitted to share confidential information. Our initial agreement runs for eight weeks, and we intend to give METR as much time as it deems necessary to complete a thorough investigation. anthropic.com/research/align…
Watched @GustavoDudamel inaugural show with the @nyphil orchestra 🫡. Impossible not to be energized by so much talent, beauty and humanity.