vedanta
Archive / long-form notesBLOG.SEQ

SECTOR_06

Research archive

Essays on agent behavior, model deployment, graph learning, and the practical edges of machine intelligence.

Research, experiments, and deployment notes from the AI systems edge.

Focused on alignment, agent behavior, graph learning, and the gap between demos and real products.

02

Graph Neural Networks Meet Quantitative Trading

A multi-relational GNN setup that blends stock features with vendor-customer relationships to treat supply-chain structure as predictive signal.

Read full post
01

Agents Are Corrupt

Exploring how autonomous agents in multi-agent environments drift toward corrupt or misaligned behavior without explicit instructions to do so.

Read full post

SECTOR_07

About

I am a software engineer and researcher interested in the point where AI research has to become a working system. That includes alignment questions, multi-agent behavior, evaluation, and the engineering needed to make those ideas useful.