Why AI Drug Discovery Needs Labs That Never Sleep

Manohar Paluri, co-founder and CTO of Warp Sciences · The Neon Show

Manohar Paluri is co-founder and CTO of Warp Sciences, a new company automating chemistry wet labs to speed drug discovery. He spent about 14 years at Meta, rising from intern to vice president, and worked closely with Yann LeCun and Mark Zuckerberg on AI. He studied at IIIT Hyderabad, worked at SRI International and earned a PhD at Georgia Tech. His co-founders include former OpenAI product chief Kevin Weil, chief science officer Saul Rosenberg and DJ Patil as COO.

Summary

Paluri says AI improved so fast at code and math because results there can be verified automatically, while the physical world is slowed by human labor, materials and long lead times. Warp wants to build robotic wet labs that run around the clock so drug discovery becomes data-rich and each experiment makes the next faster. He believes open ecosystems win over time though closed models will likely lead on capability, and that national AI efforts work only when they aim to innovate. His career rule is to spot a bottleneck, learn from experts, take the leap and measure success by real-world use. These signals are from an interview that Siddhartha Ahluwalia did with Manohar Paluri on The Neon Show.

1. The Growing Nucleus. Paluri left Meta because, he says, the problem excites him more than the position. Frontier models are making unprecedented progress, and what they can do out of the box keeps expanding. Builders can still add value with better harnesses, data and customization, and value will flow to both models and applications. He wanted to work on what pushes the models further. ▶ 3:25
   
   Quote: “It is less about being at the frontier and saying no to it and more about what is the fundamental bottleneck in learning.” (Manohar Paluri, ▶ 5:36)