AI research practiceBengaluru, India
I’m Shivam Pandey, an AI Research Engineer working across embodied intelligence, foundation models, interpretability, and the high-performance systems that make ambitious experiments possible.
Move through the field. Local trajectories reorganise around your intervention.
00 / Research thesis
My work connects the mathematical foundations of learning, the behaviour that emerges from them, and the systems required to test both. The useful questions live in the hand-offs: objective to representation, representation to action, and action to consequence.
01 / Research field
Active direction
How reward, preference, and verifiable feedback shape what a policy actually learns—and where intended behaviour separates from observed action.
Active direction
How latent world-action models encode possible futures, and whether a policy’s decisions depend on information that survives intervention and distribution shift.
Established practice
The kernels, distributed systems, and evaluation infrastructure that shorten the distance between a mathematical question and a trustworthy empirical answer.
02 / Published material
KERNELIZED / 06
KERNELIZED / 05
KERNELIZED / 04
A publication by Shivam Pandey
Deeper essays, experiments, implementations, and research series—published once, connected back to the personal research practice.
Enter the publication