Blog

Sharing my learning notes, technical insights, and research explorations

All Notes 6 items

Core notes on self-attention, token mixing, positional encoding, and the encoder-decoder ideas behind modern foundation models.

A visual primer on point clouds, meshes, voxels, implicit fields, and how different 3D representations shape perception pipelines.

Notes on recovering geometry from images and depth signals, with a focus on reconstruction assumptions, losses, and evaluation.

A compact survey of vision-language-action models, from perception-language alignment to action prediction for embodied agents.

A reading note on world models, latent dynamics, planning, and how learned simulators support decision-making.

Foundational reinforcement learning notes covering policies, value functions, exploration, and optimization loops.