Vinod Anbalagan

Writing · The Meta Gradient

From code to consciousness.

Technical essays with the philosophical bits left in. Each tries to build a visual intuition alongside the formal version.

Subscribe on Substack →

Learning & world models

  1. Freezing the NoiseHow one line of algebra made sampling differentiable.18 Sept 2026
  2. The Architecture of DreamsPart 1: how neural networks learned to imagine. World Models, PlaNet, Dreamer.14 Sept 2026
  3. Finding the ManifoldWhere neural networks land: the geometry of optimisation, with a live demo.6 Jan 2026

Data & evaluation

  1. The BreakthroughWhat 10 Grade B runs taught me about data quality.18 Apr 2026
  2. Building the Dataset Tamil Farmers DeserveWhy brute-force AI fails the Global South.9 Apr 2026
  3. On Building Offline AI for People Who Actually Need ItWhat 9,000 evaluation pairs taught us about multilingual retrieval.1 Apr 2026

Perspectives

  1. If AI Will Shape the World, the World Must Help Govern ItOpen weights, legitimacy, and who gets to participate in science.1 Aug 2026
  2. NeurIPS 2025: The Year We Admitted We Were LostAnd found a better map. Notes on the conference that challenged scaling orthodoxy.15 Dec 2025
  3. Animals, Ghosts, and ChimerasThe AI debate that will define the next decade.24 Oct 2025