Essays

The Orchestra Thesis

Perplexity Computer bets that the future of AI is not one model to rule them all, but an orchestration layer that routes work to 19 specialized models. Here's why that thesis is more compelling than the skeptics think.

Essays

The New American Playbook

American special forces captured Maduro. Oil companies are circling 303 billion barrels of reserves. And the next chapter, likely involving Iran, could reshape global energy markets entirely.

Essays

The Space Lobster Problem

OpenClaw, an open-source AI agent, went from zero to 200,000 GitHub stars in two weeks, surpassing Linux and Kubernetes. It is also a massive security liability. Here is what it means, why it matters, and why I think a major AI company is about to make a move on it.

Essays

Don't Touch the Fed

Trump just nominated Kevin Warsh to lead the Fed. Here's why the independence of the central bank matters more than ever, and what history says happens when politicians take the wheel.

Papers2024

Scaling Distributed Machine Learning Systems: A Survey

PDF

Aditya Mehrotra, Alice Johnson

arXiv preprint • 2024

A comprehensive survey of techniques for scaling machine learning systems across distributed computing environments, covering both data and model parallelism approaches.

Papers2024

Efficient Parameter-Efficient Fine-Tuning for Large Language Models

PDF

Aditya Mehrotra, John Smith, Jane Doe

NeurIPS 2024 • 2024

We propose a novel method for parameter-efficient fine-tuning of large language models that reduces memory requirements by 70% while maintaining comparable performance to full fine-tuning.

Papers2023

Adversarial Robustness in Computer Vision Models

PDF

Bob Williams, Aditya Mehrotra, Carol Davis

CVPR 2023 • 2023

We investigate adversarial attack vectors in state-of-the-art computer vision models and propose a new defense mechanism that improves robustness by 35%.