Aryan Kumar Singh

I am an MSc AI student in Berlin. I research embodied AI: robots that see, decide and act. For my thesis I train reinforcement learning policies for robot control in MuJoCo and ROS 2. At Barycenter Lab I lead the vision work on Kiko, a home robot. My second field is multimodal health AI. I test my methods with statistics before I trust them.

Research focus

Embodied AI and robot learning

Policies that learn in simulation and run on real robots.

Multimodal health AI

Models that read brain, face, voice and text signals and explain what they found.

ML evaluation

Benchmarks and statistical tests that show when a method wins.

Now (October 2026)

  • Building Kiko's vision on an NVIDIA Jetson at Barycenter Lab
  • Training RL policies for robot control for my MSc thesis
  • Building NeuroPrism and preparing the code release
  • Finishing the forecast combination paper

News

  • Oct 2026 Kiko's vision runs on the Jetson GPU, and Kiko spoke its first words.
  • Oct 2026 Joined Barycenter Lab as AI and Robotics Researcher.
  • Oct 2026 Started NeuroPrism, an explainable brain health AI project.
  • Oct 2026 Released SpotZero as open source.
  • Sep 2026 Released the code for my forecast combination paper.
  • Aug 2026 Started teaching a live robotics course.

Selected work

Kiko

A home robot that knows its owner, follows them, talks and films reels. I lead the AI and computer vision work.

Status: Hardware build in progress

NeuroPrism

Explainable brain health AI from EEG, face, speech and text. Research project. Not a medical device.

Status: In progress, code release planned

Convex forecast combination

A benchmark of 23 forecasting models on 5 energy datasets, with statistical tests and KKT boundary theory.

Status: Manuscript in preparation. Code is public.

I am looking for PhD positions and research roles in embodied AI. Email me at aksingh5114 [at] gmail [dot] com.