AI R&D at @cribl_io | Hiring researchers & developers | Purpose built models for observability, telemetry, and security
San Francisco
Joined December 2008
- Tweets820
- Following301
- Followers481
- Likes1.5K
Plan to split the benchmark into a publicly available and private held out set so that we can compare the delta across both and control for benchmaxxing
I wonder if it's because they've invested in fine-tuning harnesses or if they believe frontier intelligence is plateauing
Nikhil Mungel retweeted
Your robot arm is probably already supported by LeRobot. You just don't know it yet.
The community has shipped 30+ hardware integrations: xArm, UR5e, Franka, Trossen, AgileX Piper, I2RT YAM, ARX5, ugo Pro, Quest and PICO 4 teleop, GELLO, Haply Inverse3 haptics, ROS 2 bridges.
Most of them are drop-in plugins.
pip install
pass --robot.type=...
and you're recording datasets. No fork, no patching our repo.
huggingface.co/docs/lerobot/…
"It's just two APIs: 𝚜𝚎𝚗𝚍 and 𝚑𝚊𝚗𝚍𝚕𝚎𝙲𝚊𝚕𝚕𝚋𝚊𝚌𝚔" I love how simple yet elegant vercel's APIs turn out, it's like the opposite of aws
Nikhil Mungel retweeted
#AI promises total automation for security investigations...
...but the reality is that teams are hitting a wall: investigations remain agonizingly slow.
The failure isn't the AI, it’s the data feeding the machine.
Get your data AI-ready with @nmungel: bit.ly/3M1YXAN
Nikhil Mungel retweeted
🎆 New year, new #devops enginner! We've got your perfect resolution to help you fully embrace #AI in 2026: Develop the skills required to use AI for solving development and engineering challenges: bit.ly/45rO8yg
@InfoWorld
@nmungel
Nikhil Mungel retweeted
Having some AI follow you into your zoom meetings or google meet for taking notes is the digital equivalent of showing up to a meeting with your fly down
I think this will take off once there's a way to bake the learning into the model's weights at least daily. If the learning is simply clever context engineering then I'm less optimistic -- maybe it works for narrow tasks, but hard to see how it'll scale generally, wdyt?
Nikhil Mungel retweeted
With $FIG we have another massive "POP" highlighting the gross inefficiency in the modern IPO process. It's very simple. They REFUSE to match supply/demand (that happened today). They brag about the mis-match - "30X oversubscribed." The outcome is expected & fully intentional.
