Parv Kapoor

Last year, I came across the idea of constrained decoding (I know, late to the party) and was fascinated. The ability to enforce constraints for LLMs at inference time without fine tuning is a powerful idea. It got me thinking, can we do this for robot foundation models? 1/n🧵

There is a poetic depth to the term "latent space" that transforms vector coordinates into a frontier of pure possibility

waiting for ICLR to delay the decisions further due to the snowstorm

always in awe of this city

skill issue

@JohnCLangford will be presenting our recent work on a principled way to learn compact world models in transformers. If you’re at NeurIPS, please do check out the workshop! 😊

@pratyusha_PS @pratyushmaini Congratulations @pratyusha_PS , it was a delight interacting with you over the summer!

📢 Some big (& slightly belated) life updates! 1. I defended my PhD at MIT this summer! 🎓 2. I'm joining NYU as an Assistant Professor starting Fall 2026, with a joint appointment in Courant CS and the Center for Data Science. 🎉 🔬 My lab will focus on empirically studying the science of deep learning and applying deep learning to accelerate the natural sciences. Very broadly interested in questions at the intersection of language, reasoning and sequential decision making. (Plus any other fun problems that catch our eye along the way!) 🚀 I am recruiting 2 PhD students for this cycle! If you're interested in joining, please apply here: cs.nyu.edu/dynamic/phd/ad… cds.nyu.edu/phd-admissions…

someone needs to start a polymarket on openreview crashing on review release day and i’ll go all in on it

Is it just me or did they use a thee sacred souls instrumental for the intro background music?? (the will i see you again song) @1x_tech

How can we create a single navigation policy that works for different robots in diverse environments AND can reach navigation goals with high precision? Happy to share our new paper, "VAMOS: A Hierarchical Vision-Language-Action Model for Capability-Modulated and Steerable Navigation"! 📜 Paper: arxiv.org/abs/2510.20818 🌐 Website: vamos-vla.github.io

@mangahomanga @JohnsHopkins @JHUCompSci Congrats!

I'll be joining the faculty @JohnsHopkins late next year as a tenure-track assistant professor in @JHUCompSci Looking for PhD students to join me tackling fun problems in robot manipulation, learning from human data, understanding+predicting physical interactions, and beyond!

@chaitanya1cha thanks for the kind words! that is actually a very interesting direction, i might just look at this next 👀

Really interesting application of constrained decoding! I think this could be scaled to test-time robot-embodiment adaptation as well, where VLA outputs are constrained to predict actions based on the current embodiment configurations.

LLMs got JSON constraints, now RFMs get safety constraints. With SafeDec, you can plug in contextual rules and make your model behave safely. Read our full blog here 👇 parvkpr.github.io/projects/safed… n/n🧵

Evaluated on hundreds of procedurally generated AI2-THOR scenes with three SOTA policies (SPOC, Flare, PoliFormer), SafeDec enforced two invariant specs with 0 violations: avoid (never enter forbidden zones) and geofence (stay within allowed regions). 8/n🧵

Last year, I came across the idea of constrained decoding (I know, late to the party) and was fascinated. The ability to enforce constraints for LLMs at inference time without fine tuning is a powerful idea. It got me thinking, can we do this for robot foundation models? 1/n🧵