Prabakaran Chandran

பிரபாகரன் சந்திரன்

Senior ML Engineer · MS Data Science, Columbia University (2025–2027)


About

ML Engineer and researcher with 6.5+ years building and deploying systems across document understanding, enterprise AI, forecasting and time-series modeling, computer vision, optimization, and simulation. Currently pursuing an MS in Data Science at Columbia University, with a focus on the theoretical and applied foundations of learning and decision-making. Teaching Assistant in Causal Inference and Advanced Analytics.

The work sits at the intersection of Systems, Behavior, and Intelligence — how dynamical systems evolve, how people and organizations actually decide, and how machine learning operates inside both. The organizing question is not prediction but intervention: what changes when you act, and how does the system respond? This framing — decision engineering — shapes the research, the tools, and the writing.


Research Interests

  • Causal inference and invariance
  • Reinforcement learning and multi-agent systems
  • Dynamical systems and time-series modeling
  • Continual learning and memory
  • Interpretability via simulation and counterfactual reasoning
  • Mechanism design and behavioral modeling

Currently — March 2026

  • Study MS Data Science at Columbia — RL, causal inference, deep learning, probabilistic modeling
  • Teaching TA in Causal Inference and Advanced Analytics
  • Writing cooc.ing Vol. I — first essay live, next in progress
  • Research Generative model experiments at appliedmodels.co; mathematical foundations at researchengineer.ing
  • Building broCoDDE live at prachalabs.com; Rewire and Bounded Agents in development

On This Site

  • Thinking — Essays and arguments at the intersection of AI, cognition, and complex systems.
  • Doing — Research notes, reading logs, and experiments in progress.
  • Living — Manhattan, Tamil Nadu, and what happens between the work.

Elsewhere

  • Labs — Product lab and independent research studio.
  • Daily — Reflections and working notes.

Connect