Nishanth Kumar

@nishanthkumar23

Research @Meta Robotics Studio. Ph.D. @MIT_LISLab. Formerly @AIatMeta., @NVIDIAAI, @rai_inst, @vicariousai and @uber; Sc.B. @BrownUniversity.

Cambridge, MA

nishanthjkumar.com

Replies Reposts Media Articles

Pinned
Nishanth Kumar
@nishanthkumar23
Jun 3
Life update: after finishing up my PhD at @MIT_LISLab @MIT_CSAIL, I'm excited to join the Robotics Studio team at @Meta as a Research Scientist! I feel lucky to get to work with a team full of exceptional colleagues committed to pushing the boundaries of intelligent robots. P.S:
13
Nishanth Kumar
@nishanthkumar23
4h
As promising and exciting as “robot-use” agents are, there aren’t a lot of good frameworks to develop the necessary harnesses and coordinate robot tools. Retriever seems like a great step in this direction, and I suspect more work like it will be increasingly important!
Linfeng Zhao
@LinfengZhaoZLF
Sep 15
How should we build long-horizon manipulation systems? As coding agents become more capable, what interface should connect models to the physical world? We’ve been thinking about this since 2024. Introducing Retriever, a programming framework for asynchronous, closed-loop robot
00:00
2
Nishanth Kumar
@nishanthkumar23
Aug 21
It was super fun to get to show a glimpse into our robotics work as part of this release: research.meta.ai/blog/multimoda… It's a very exciting time to be working on robotics @Meta - stay tuned for more! 🤖
Alexandr Wang
@alexandr_wang
Aug 20
1/ muse spark 1.2 is a very strong multimodal model—it can do visual coding, robotics planning, and audio-visual understanding that all come together through agentic tools.
00:00
00:00
2
Nishanth Kumar
@nishanthkumar23
Jun 25
Lots of great new posts on the AI/ML industry job market + interviewing graduating PhDs! Posts like these really helped me, so I figured I'd contribute to the genre myself. Here's a blog post based on my experiences interviewing in robotics: nishanthjkumar.com/blog/2026/Navi…
4
Nishanth Kumar
@nishanthkumar23
May 8
We ran TiPToP on MolmoSpaces and achieved SOTA! And we wrote an X article thing to discuss details - excited to hear any questions/comments! Check out the leaderboard here: molmospaces.allen.ai/leaderboard. TiPToP is also fully-open and should take <1hr of setup: try it yourself here:
Nishanth Kumar
@nishanthkumar23
May 8
Article
Achieving SOTA on the MolmoSpaces benchmark with Inference-Time Search
We recently developed TiPToP: a general-purpose manipulation system that uses foundation models for perception followed by inference-time search for trajectory generation (specifically via Task and...
1